seqsero2s: SeqSero2S/bin/SeqSero2S.py annotate

author	jpayne
date	Fri, 15 May 2026 17:50:45 +0000
parents
children

rev	line source
19 cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	1 #!/usr/bin/env python3
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	2
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	3 import sys
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	4 import time
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	5 import random
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	6 import os
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	7 import subprocess
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	8 import gzip
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	9 import io
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	10 import pickle
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	11 import argparse
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	12 import itertools
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	13 import json
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	14 from distutils.version import LooseVersion
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	15 from distutils.spawn import find_executable
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	16 sys.path.insert(1,sys.path[0]+'/..')
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	17
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	18 __version__ = "1.1.4"
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	19
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	20 ### SeqSero Kmer
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	21 def parse_args():
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	22 "Parse the input arguments, use '-h' for help."
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	23 parser = argparse.ArgumentParser(usage='SeqSero2S.py -t <data_type> -m <mode> -i <input_data> [-d <output_directory>] [-p <number of threads>] [-b <BWA_algorithm>]\n\nDevelopper: Shaokang Zhang (zskzsk@uga.edu), Hendrik C Den-Bakker (Hendrik.DenBakker@uga.edu) and Xiangyu Deng (xdeng@uga.edu)\n\nContact email:seqsero@gmail.com\n\n')#add "-m <data_type>" in future
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	24 parser.add_argument("-i",nargs="+",help="<string>: path/to/input_data",type=os.path.abspath) ### add 'type=os.path.abspath' to generate absolute path of input data.
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	25 parser.add_argument("-t",choices=['1','2','3','4','5'],help="<int>: '1' for interleaved paired-end reads, '2' for separated paired-end reads, '3' for single reads, '4' for genome assembly, '5' for nanopore reads (fasta/fastq)")
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	26 parser.add_argument("-b",choices=['sam','mem'],default="mem",help="<string>: algorithms for bwa mapping for allele mode; 'mem' for mem, 'sam' for samse/sampe; default=mem; optional; for now we only optimized for default 'mem' mode")
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	27 parser.add_argument("-p",default="1",help="<int>: number of threads for allele mode, if p >4, only 4 threads will be used for assembly since the amount of extracted reads is small, default=1")
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	28 parser.add_argument("-m",choices=['k','a'],default="a",help="<string>: which workflow to apply, 'a'(raw reads allele micro-assembly), 'k'(raw reads and genome assembly k-mer), default=a")
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	29 parser.add_argument("-n",help="<string>: optional, to specify a sample name in the report output")
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	30 parser.add_argument("-d",help="<string>: optional, to specify an output directory name, if not set, the output directory would be 'SeqSero_result_'+time stamp+one random number")
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	31 parser.add_argument("-c",action="store_true",help="<flag>: if '-c' was flagged, SeqSero2S will only output serotype prediction without the directory containing log files")
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	32 parser.add_argument("-s",action="store_true",help="<flag>: if '-s' was flagged, SeqSero2S will not output header in SeqSero_result.tsv")
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	33 parser.add_argument("--phred_offset",choices=['33','64','auto'],default='auto',help="<33\|64\|auto>: offset for FASTQ file quality scores, default=auto")
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	34 parser.add_argument("--check",action="store_true",help="<flag>: use '--check' flag to check the required dependencies")
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	35 parser.add_argument('-v', '--version', action='version', version=f"%(prog)s {__version__}")
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	36 return parser.parse_args()
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	37
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	38 ### check paths of dependencies
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	39 check_dependencies = parse_args().check
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	40 dependencies = ['bwa','samtools','blastn','fastq-dump','spades.py','bedtools','SalmID.py','mlst','stringMLST.py']
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	41 if check_dependencies:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	42 for item in dependencies:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	43 ext_path = find_executable(item)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	44 if ext_path is not None:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	45 print ("Using "+item+" - "+ext_path)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	46 else:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	47 print ("ERROR: can not find "+item+" in PATH")
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	48 sys.exit()
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	49 ### end of --check
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	50
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	51 def reverse_complement(sequence):
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	52 complement = {
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	53 'A': 'T',
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	54 'C': 'G',
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	55 'G': 'C',
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	56 'T': 'A',
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	57 'N': 'N',
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	58 'M': 'K',
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	59 'R': 'Y',
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	60 'W': 'W',
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	61 'S': 'S',
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	62 'Y': 'R',
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	63 'K': 'M',
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	64 'V': 'B',
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	65 'H': 'D',
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	66 'D': 'H',
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	67 'B': 'V'
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	68 }
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	69 return "".join(complement[base] for base in reversed(sequence))
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	70
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	71 def mlst(assembly):
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	72 subprocess.check_call("mlst -q --json mlst.json --scheme senterica_achtman_2 "+assembly+" >> data_log.txt 2>&1",shell=True)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	73 f = open("mlst.json",'r')
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	74 mlst_result = json.load(f)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	75 f.close()
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	76 st = mlst_result[0]['sequence_type']
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	77 alleles = mlst_result[0]['alleles']
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	78 return(st,alleles)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	79
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	80 def stringmlst(f1,f2,t,d):
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	81 if t in ['1','2']:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	82 subprocess.check_call("stringMLST.py --predict -P "+d+"/kmer/salmonella -1 "+f1+" -2 "+f2+" -a /dev/stdout -o stringMLST.txt >> data_log.txt 2>&1",shell=True)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	83 elif t=='3':
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	84 subprocess.check_call("stringMLST.py --predict -P "+d+"/kmer/salmonella -1 "+f1+" -s -a /dev/stdout -o stringMLST.txt >> data_log.txt 2>&1",shell=True)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	85 f = "stringMLST.txt"
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	86 mlst_result = open(f).readlines()[1].strip().split('\t')
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	87 st = mlst_result[-1]
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	88 if st == '0':
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	89 st = '-'
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	90 k = ['aroC','dnaN','hemD','hisD','purE','sucA','thrA']
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	91 v = [mlst_result[1],mlst_result[2],mlst_result[3],mlst_result[4],mlst_result[5],mlst_result[6],mlst_result[7]]
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	92 alleles = dict(zip(k,v))
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	93 return(st,alleles)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	94
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	95 def createKmerDict_reads(list_of_strings, kmer):
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	96 kmer_table = {}
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	97 for string in list_of_strings:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	98 sequence = string.strip('\n')
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	99 for i in range(len(sequence) - kmer + 1):
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	100 new_mer = sequence[i:i + kmer].upper()
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	101 new_mer_rc = reverse_complement(new_mer)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	102 if new_mer in kmer_table:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	103 kmer_table[new_mer.upper()] += 1
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	104 else:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	105 kmer_table[new_mer.upper()] = 1
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	106 if new_mer_rc in kmer_table:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	107 kmer_table[new_mer_rc.upper()] += 1
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	108 else:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	109 kmer_table[new_mer_rc.upper()] = 1
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	110 return kmer_table
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	111
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	112
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	113 def multifasta_dict(multifasta):
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	114 multifasta_list = [
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	115 line.strip() for line in open(multifasta, 'r') if len(line.strip()) > 0
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	116 ]
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	117 headers = [i for i in multifasta_list if i[0] == '>']
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	118 multifasta_dict = {}
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	119 for h in headers:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	120 start = multifasta_list.index(h)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	121 for element in multifasta_list[start + 1:]:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	122 if element[0] == '>':
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	123 break
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	124 else:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	125 if h[1:] in multifasta_dict:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	126 multifasta_dict[h[1:]] += element
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	127 else:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	128 multifasta_dict[h[1:]] = element
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	129 return multifasta_dict
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	130
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	131
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	132 def multifasta_single_string(multifasta):
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	133 multifasta_list = [
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	134 line.strip() for line in open(multifasta, 'r')
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	135 if (len(line.strip()) > 0) and (line.strip()[0] != '>')
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	136 ]
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	137 return ''.join(multifasta_list)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	138
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	139
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	140 def chunk_a_long_sequence(long_sequence, chunk_size=60):
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	141 chunk_list = []
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	142 steps = len(long_sequence) // 60 #how many chunks
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	143 for i in range(steps):
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	144 chunk_list.append(long_sequence[i * chunk_size:(i + 1) * chunk_size])
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	145 chunk_list.append(long_sequence[steps * chunk_size:len(long_sequence)])
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	146 return chunk_list
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	147
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	148
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	149 def target_multifasta_kmerizer(multifasta, k, kmerDict):
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	150 forward_length = 300 #if find the target, put forward 300 bases
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	151 reverse_length = 2200 #if find the target, put backward 2200 bases
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	152 chunk_size = 60 #it will firstly chunk the single long sequence to multiple smaller sequences, it controls the size of those smaller sequences
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	153 target_mers = []
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	154 long_single_string = multifasta_single_string(multifasta)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	155 multifasta_list = chunk_a_long_sequence(long_single_string, chunk_size)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	156 unit_length = len(multifasta_list[0])
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	157 forward_lines = int(forward_length / unit_length) + 1
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	158 reverse_lines = int(forward_length / unit_length) + 1
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	159 start_num = 0
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	160 end_num = 0
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	161 for i in range(len(multifasta_list)):
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	162 if i not in range(start_num, end_num): #avoid computational repetition
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	163 line = multifasta_list[i]
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	164 start = int((len(line) - k) // 2)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	165 s1 = line[start:k + start]
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	166 if s1 in kmerDict: #detect it is a potential read or not (use the middle part)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	167 if i - forward_lines >= 0:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	168 start_num = i - forward_lines
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	169 else:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	170 start_num = 0
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	171 if i + reverse_lines <= len(multifasta_list) - 1:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	172 end_num = i + reverse_lines
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	173 else:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	174 end_num = len(multifasta_list) - 1
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	175 target_list = [
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	176 x.strip() for x in multifasta_list[start_num:end_num]
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	177 ]
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	178 target_line = "".join(target_list)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	179 target_mers += [
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	180 k1 for k1 in createKmerDict_reads([str(target_line)], k)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	181 ] ##changed k to k1, just want to avoid the mixes of this "k" (kmer) to the "k" above (kmer length)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	182 else:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	183 pass
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	184 return set(target_mers)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	185
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	186
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	187 def target_read_kmerizer(file, k, kmerDict):
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	188 i = 1
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	189 n_reads = 0
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	190 total_coverage = 0
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	191 target_mers = []
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	192 if file.endswith(".gz"):
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	193 file_content = io.BufferedReader(gzip.open(file))
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	194 else:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	195 file_content = open(file, "r").readlines()
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	196 for line in file_content:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	197 start = int((len(line) - k) // 2)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	198 if i % 4 == 2:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	199 if file.endswith(".gz"):
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	200 s1 = line[start:k + start].decode()
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	201 line = line.decode()
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	202 else:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	203 s1 = line[start:k + start]
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	204 if s1 in kmerDict: #detect it is a potential read or not (use the middle part)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	205 n_reads += 1
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	206 total_coverage += len(line)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	207 target_mers += [
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	208 k1 for k1 in createKmerDict_reads([str(line)], k)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	209 ] #changed k to k1, just want to avoid the mixes of this "k" (kmer) to the "k" above (kmer length)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	210 i += 1
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	211 if total_coverage >= 4000000:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	212 break
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	213 return set(target_mers)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	214
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	215
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	216 def minion_fasta_kmerizer(file, k, kmerDict):
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	217 i = 1
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	218 n_reads = 0
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	219 total_coverage = 0
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	220 target_mers = {}
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	221 for line in open(file):
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	222 if i % 2 == 0:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	223 for kmer, rc_kmer in kmers(line.strip().upper(), k):
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	224 if (kmer in kmerDict) or (rc_kmer in kmerDict):
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	225 if kmer in target_mers:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	226 target_mers[kmer] += 1
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	227 else:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	228 target_mers[kmer] = 1
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	229 if rc_kmer in target_mers:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	230 target_mers[rc_kmer] += 1
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	231 else:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	232 target_mers[rc_kmer] = 1
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	233 i += 1
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	234 return set([h for h in target_mers])
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	235
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	236
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	237 def minion_fastq_kmerizer(file, k, kmerDict):
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	238 i = 1
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	239 n_reads = 0
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	240 total_coverage = 0
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	241 target_mers = {}
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	242 for line in open(file):
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	243 if i % 4 == 2:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	244 for kmer, rc_kmer in kmers(line.strip().upper(), k):
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	245 if (kmer in kmerDict) or (rc_kmer in kmerDict):
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	246 if kmer in target_mers:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	247 target_mers[kmer] += 1
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	248 else:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	249 target_mers[kmer] = 1
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	250 if rc_kmer in target_mers:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	251 target_mers[rc_kmer] += 1
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	252 else:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	253 target_mers[rc_kmer] = 1
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	254 i += 1
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	255 return set([h for h in target_mers])
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	256
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	257
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	258 def multifasta_single_string2(multifasta):
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	259 single_string = ''
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	260 with open(multifasta, 'r') as f:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	261 for line in f:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	262 if line.strip()[0] == '>':
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	263 pass
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	264 else:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	265 single_string += line.strip()
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	266 return single_string
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	267
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	268
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	269 def kmers(seq, k):
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	270 rev_comp = reverse_complement(seq)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	271 for start in range(1, len(seq) - k + 1):
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	272 yield seq[start:start + k], rev_comp[-(start + k):-start]
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	273
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	274
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	275 def multifasta_to_kmers_dict(multifasta,k_size):#used to create database kmer set
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	276 multi_seq_dict = multifasta_dict(multifasta)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	277 lib_dict = {}
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	278 for h in multi_seq_dict:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	279 lib_dict[h] = set(
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	280 [k for k in createKmerDict_reads([multi_seq_dict[h]], k_size)])
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	281 return lib_dict
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	282
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	283
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	284 def Combine(b, c):
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	285 fliC_combinations = []
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	286 fliC_combinations.append(",".join(c))
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	287 temp_combinations = []
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	288 for i in range(len(b)):
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	289 for x in itertools.combinations(b, i + 1):
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	290 temp_combinations.append(",".join(x))
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	291 for x in temp_combinations:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	292 temp = []
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	293 for y in c:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	294 temp.append(y)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	295 temp.append(x)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	296 temp = ",".join(temp)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	297 temp = temp.split(",")
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	298 temp.sort()
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	299 temp = ",".join(temp)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	300 fliC_combinations.append(temp)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	301 return fliC_combinations
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	302
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	303
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	304 def seqsero_from_formula_to_serotypes(Otype, fliC, fljB, special_gene_list,subspecies,ss):
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	305 #like test_output_06012017.txt
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	306 #can add more varialbles like sdf-type, sub-species-type in future (we can conclude it into a special-gene-list)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	307 #from Initial_Conditions import phase1,phase2,phaseO,sero,subs,remove_list,rename_dict
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	308 if ss == 'ss2':
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	309 from Initial_Conditions_SS2 import phase1,phase2,phaseO,sero,subs,remove_list,rename_dict
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	310 elif ss == 'ss2s':
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	311 from Initial_Conditions_SS2S import phase1,phase2,phaseO,sero,subs,remove_list,rename_dict
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	312 rename_dict_not_anymore=[rename_dict[x] for x in rename_dict]
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	313 rename_dict_all=rename_dict_not_anymore+list(rename_dict) #used for decide whether to
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	314 seronames = []
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	315 seronames_none_subspecies=[]
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	316 for i in range(len(phase1)):
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	317 fliC_combine = []
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	318 fljB_combine = []
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	319 if phaseO[i] == Otype: # no VII in KW, but it's there
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	320 ### for fliC, detect every possible combinations to avoid the effect of "["
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	321 if phase1[i].count("[") == 0:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	322 fliC_combine.append(phase1[i])
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	323 elif phase1[i].count("[") >= 1:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	324 c = []
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	325 b = []
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	326 if phase1[i][0] == "[" and phase1[i][-1] == "]" and phase1[i].count(
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	327 "[") == 1:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	328 content = phase1[i].replace("[", "").replace("]", "")
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	329 fliC_combine.append(content)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	330 fliC_combine.append("-")
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	331 else:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	332 for x in phase1[i].split(","):
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	333 if "[" in x:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	334 b.append(x.replace("[", "").replace("]", ""))
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	335 else:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	336 c.append(x)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	337 fliC_combine = Combine(
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	338 b, c
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	339 ) #Combine will offer every possible combinations of the formula, like f,[g],t: f,t f,g,t
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	340 ### end of fliC "[" detect
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	341 ### for fljB, detect every possible combinations to avoid the effect of "["
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	342 if phase2[i].count("[") == 0:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	343 fljB_combine.append(phase2[i])
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	344 elif phase2[i].count("[") >= 1:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	345 d = []
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	346 e = []
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	347 if phase2[i][0] == "[" and phase2[i][-1] == "]" and phase2[i].count(
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	348 "[") == 1:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	349 content = phase2[i].replace("[", "").replace("]", "")
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	350 fljB_combine.append(content)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	351 fljB_combine.append("-")
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	352 else:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	353 for x in phase2[i].split(","):
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	354 if "[" in x:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	355 d.append(x.replace("[", "").replace("]", ""))
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	356 else:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	357 e.append(x)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	358 fljB_combine = Combine(d, e)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	359 ### end of fljB "[" detect
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	360 new_fliC = fliC.split(
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	361 ","
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	362 ) #because some antigen like r,[i] not follow alphabetical order, so use this one to judge and can avoid missings
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	363 new_fliC.sort()
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	364 new_fliC = ",".join(new_fliC)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	365 new_fljB = fljB.split(",")
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	366 new_fljB.sort()
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	367 new_fljB = ",".join(new_fljB)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	368 if (new_fliC in fliC_combine
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	369 or fliC in fliC_combine) and (new_fljB in fljB_combine
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	370 or fljB in fljB_combine):
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	371 ######start, remove_list,rename_dict, added on 11/11/2018
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	372 if sero[i] not in remove_list:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	373 temp_sero=sero[i]
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	374 if temp_sero in rename_dict:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	375 temp_sero=rename_dict[temp_sero] #rename if in the rename list
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	376 if temp_sero not in seronames:#the new sero may already included, if yes, then not consider
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	377 if subs[i] == subspecies:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	378 seronames.append(temp_sero)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	379 seronames_none_subspecies.append(temp_sero)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	380 else:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	381 pass
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	382 else:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	383 pass
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	384 ######end, added on 11/11/2018
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	385 #analyze seronames
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	386 subspecies_pointer=""
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	387 if len(seronames) == 0 and len(seronames_none_subspecies)!=0:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	388 ## ed_SL_06062020: for the subspecies mismatch between KW and SalmID
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	389 seronames=seronames_none_subspecies
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	390 #seronames=["N/A"]
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	391 subspecies_pointer="1"
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	392 #subspecies_pointer="0"
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	393 ##
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	394 if len(seronames) == 0:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	395 seronames = [
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	396 "N/A (The predicted antigenic profile does not exist in the White-Kauffmann-Le Minor scheme)"
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	397 ]
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	398 star = ""
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	399 star_line = ""
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	400 if len(seronames) > 1: #there are two possible predictions for serotypes
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	401 star = "*"
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	402 #changed 04072019
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	403 #star_line = "The predicted serotypes share the same general formula:\t" + Otype + ":" + fliC + ":" + fljB + "\n"
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	404 if subspecies_pointer=="1" and len(seronames_none_subspecies)!=0:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	405 star="*"
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	406 star_line = "This antigenic profile has been associated with serotype '"+(" or ").join(seronames)+"' in the Kauffman-White scheme. The existence of the same antigenic formula in multiple species or subspecies is well documented in the Kauffman-White Scheme. " + star_line ## ed_SL_03202021: changed for new output format
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	407 #star_line="The predicted O and H antigens correspond to serotype '"+(" or ").join(seronames)+"' in the Kauffmann-White scheme. The predicted subspecies by SalmID (github.com/hcdenbakker/SalmID) may not be consistent with subspecies designation in the Kauffmann-White scheme. " + star_line
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	408 #star_line="The formula with this subspieces prediction can't get a serotype in KW manual, and the serotyping prediction was made without considering it."+star_line
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	409 seronames=["N/A"] ## ed_SL_06062020
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	410 if Otype=="":
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	411 Otype="-"
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	412 predict_form = Otype + ":" + fliC + ":" + fljB
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	413 predict_sero = (" or ").join(seronames)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	414 ###special test for Enteritidis
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	415 if predict_form == "9:g,m:-":
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	416 sdf = "-"
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	417 for x in special_gene_list:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	418 if x.startswith("sdf"):
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	419 sdf = "+"
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	420 #star_line="Detected sdf gene, a marker to differentiate Gallinarum and Enteritidis"
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	421 #star_line="sdf gene detected. "
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	422 star_line = "Detected Sdf I that is characteristic of commonly circulating strains of serotype Enteritidis. "
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	423 #predict_form = predict_form + " Sdf prediction:" + sdf
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	424 predict_form = predict_form #changed 04072019
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	425 if sdf == "-":
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	426 star = "*"
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	427 #star_line="Didn't detected sdf gene, a marker to differentiate Gallinarum and Enteritidis"
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	428 #star_line="sdf gene not detected. "
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	429 star_line = "Sdf I that is characteristic of commonly circulating strains of serotype Enteritidis was not detected. "
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	430 #changed in 04072019, for new output
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	431 #star_line = "Additional characterization is necessary to assign a serotype to this strain. Commonly circulating strains of serotype Enteritidis are sdf+, although sdf- strains of serotype Enteritidis are known to exist. Serotype Gallinarum is typically sdf- but should be quite rare. Sdf- strains of serotype Enteritidis and serotype Gallinarum can be differentiated by phenotypic profile or genetic criteria.\n"
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	432 #predict_sero = "Gallinarum/Enteritidis" #04132019, for new output requirement
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	433 predict_sero = "Gallinarum or Enteritidis"
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	434 ###end of special test for Enteritidis
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	435 elif predict_form == "4:i:-":
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	436 predict_sero = "I 4,[5],12:i:-" # change serotype name
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	437 elif predict_form == "4:r:-":
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	438 predict_sero = "N/A (4:r:-)"
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	439 elif predict_form == "4:b:-":
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	440 predict_sero = "N/A (4:b:-)"
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	441 #elif predict_form == "8:e,h:1,2": #removed after official merge of newport and bardo
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	442 #predict_sero = "Newport"
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	443 #star = "*"
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	444 #star_line = "Serotype Bardo shares the same antigenic profile with Newport, but Bardo is exceedingly rare."
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	445 claim = "The serotype(s) is/are the only serotype(s) with the indicated antigenic profile currently recognized in the Kauffmann White Scheme. New serotypes can emerge and the possibility exists that this antigenic profile may emerge in a different subspecies. Identification of strains to the subspecies level should accompany serotype determination; the same antigenic profile in different subspecies is considered different serotypes.\n"
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	446 if "N/A" in predict_sero:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	447 claim = ""
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	448 #special test for Typhimurium
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	449 if "Typhimurium" in predict_sero or predict_form == "4:i:-":
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	450 normal = 0
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	451 mutation = 0
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	452 for x in special_gene_list:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	453 if "oafA-O-4_full" in x:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	454 normal = float(special_gene_list[x])
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	455 elif "oafA-O-4_5-" in x:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	456 mutation = float(special_gene_list[x])
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	457 if normal > mutation:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	458 pass
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	459 elif normal < mutation:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	460 #predict_sero = predict_sero.strip() + "(O5-)"
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	461 predict_sero = predict_sero.strip() #diable special sero for new output requirement, 04132019
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	462 star = "*"
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	463 #star_line = "Detected the deletion of O5-."
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	464 star_line = "Detected a deletion in gene oafA that causes O5- variant of Typhimurium. "
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	465 else:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	466 pass
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	467 #special test for Paratyphi B
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	468 if "Paratyphi B" in predict_sero or predict_form == "4:b:-":
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	469 normal = 0
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	470 mutation = 0
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	471 for x in special_gene_list:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	472 if "gntR-family-regulatory-protein_dt-positive" in x:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	473 normal = float(special_gene_list[x])
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	474 elif "gntR-family-regulatory-protein_dt-negative" in x:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	475 mutation = float(special_gene_list[x])
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	476 #print(normal,mutation)
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	477 if normal > mutation:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	478 #predict_sero = predict_sero.strip() + "(dt+)" #diable special sero for new output requirement, 04132019
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	479 predict_sero = predict_sero.strip()+' var. L(+) tartrate+' if "Paratyphi B" in predict_sero else predict_sero.strip()
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	480 star = "*"
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	481 #star_line = "Didn't detect the SNP for dt- which means this isolate is a Paratyphi B variant L(+) tartrate(+)."
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	482 star_line = "The SNP in gene STM3356 that is associated with the d-Tartrate nonfermenting phenotype characteristic of the typhoidal pathotype was not detected. "
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	483 elif normal < mutation:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	484 #predict_sero = predict_sero.strip() + "(dt-)" #diable special sero for new output requirement, 04132019
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	485 predict_sero = predict_sero.strip()
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	486 star = "*"
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	487 #star_line = "Detected the SNP for d-Tartrate nonfermenting phenotype of Paratyphi B. "
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	488 star_line = "Detected the SNP in gene STM3356 that is associated with the d-Tartrate nonfermenting phenotype characteristic of the typhoidal pathotype. "
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	489 else:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	490 star = "*"
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	491 #star_line = " Failed to detect the SNP for dt-, can't decide it's a Paratyphi B variant L(+) tartrate(+) or not."
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	492 star_line = " " ## ed_SL_05152019: do not report this situation.
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	493 #special test for O13,22 and O13,23
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	494 ### add comment for any O2 call. 06052024
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	495 if Otype=='2' and predict_sero not in ['Nitra','Kiel','Koessen']:
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	496 star_line = 'O2 is typically a O9 rfb serotype with a mutation that results in a different sugar being placed in O antigen. SS2S detects only one group O2 serotype, Paratyphi A. This genome may be a variant of a group O9 serotype.'
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	497 ###
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	498 if Otype=="13":
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	499 #ex_dir = os.path.dirname(os.path.realpath(__file__))
cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895 jpayne parents: diff changeset	500 ex_dir = os.path.abspath(os.path.join(os.path.dirname(os.path.dirname(__file__)),'seqsero2s_db')) # ed_SL_09152019

19

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1 #!/usr/bin/env python3

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

2

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

3 import sys

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

4 import time

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

5 import random

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

6 import os

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

7 import subprocess

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

8 import gzip

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

9 import io

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

10 import pickle

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

11 import argparse

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

12 import itertools

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

13 import json

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

14 from distutils.version import LooseVersion

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

15 from distutils.spawn import find_executable

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

16 sys.path.insert(1,sys.path[0]+'/..')

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

17

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

18 __version__ = "1.1.4"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

19

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

20 ### SeqSero Kmer

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

21 def parse_args():

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

22 "Parse the input arguments, use '-h' for help."

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

23 parser = argparse.ArgumentParser(usage='SeqSero2S.py -t <data_type> -m <mode> -i <input_data> [-d <output_directory>] [-p <number of threads>] [-b <BWA_algorithm>]\n\nDevelopper: Shaokang Zhang (zskzsk@uga.edu), Hendrik C Den-Bakker (Hendrik.DenBakker@uga.edu) and Xiangyu Deng (xdeng@uga.edu)\n\nContact email:seqsero@gmail.com\n\n')#add "-m <data_type>" in future

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

24 parser.add_argument("-i",nargs="+",help="<string>: path/to/input_data",type=os.path.abspath) ### add 'type=os.path.abspath' to generate absolute path of input data.

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

25 parser.add_argument("-t",choices=['1','2','3','4','5'],help="<int>: '1' for interleaved paired-end reads, '2' for separated paired-end reads, '3' for single reads, '4' for genome assembly, '5' for nanopore reads (fasta/fastq)")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

26 parser.add_argument("-b",choices=['sam','mem'],default="mem",help="<string>: algorithms for bwa mapping for allele mode; 'mem' for mem, 'sam' for samse/sampe; default=mem; optional; for now we only optimized for default 'mem' mode")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

27 parser.add_argument("-p",default="1",help="<int>: number of threads for allele mode, if p >4, only 4 threads will be used for assembly since the amount of extracted reads is small, default=1")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

28 parser.add_argument("-m",choices=['k','a'],default="a",help="<string>: which workflow to apply, 'a'(raw reads allele micro-assembly), 'k'(raw reads and genome assembly k-mer), default=a")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

29 parser.add_argument("-n",help="<string>: optional, to specify a sample name in the report output")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

30 parser.add_argument("-d",help="<string>: optional, to specify an output directory name, if not set, the output directory would be 'SeqSero_result_'+time stamp+one random number")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

31 parser.add_argument("-c",action="store_true",help="<flag>: if '-c' was flagged, SeqSero2S will only output serotype prediction without the directory containing log files")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

32 parser.add_argument("-s",action="store_true",help="<flag>: if '-s' was flagged, SeqSero2S will not output header in SeqSero_result.tsv")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

33 parser.add_argument("--phred_offset",choices=['33','64','auto'],default='auto',help="<33|64|auto>: offset for FASTQ file quality scores, default=auto")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

34 parser.add_argument("--check",action="store_true",help="<flag>: use '--check' flag to check the required dependencies")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

35 parser.add_argument('-v', '--version', action='version', version=f"%(prog)s {__version__}")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

36 return parser.parse_args()

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

37

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

38 ### check paths of dependencies

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

39 check_dependencies = parse_args().check

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

40 dependencies = ['bwa','samtools','blastn','fastq-dump','spades.py','bedtools','SalmID.py','mlst','stringMLST.py']

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

41 if check_dependencies:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

42 for item in dependencies:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

43 ext_path = find_executable(item)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

44 if ext_path is not None:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

45 print ("Using "+item+" - "+ext_path)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

46 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

47 print ("ERROR: can not find "+item+" in PATH")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

48 sys.exit()

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

49 ### end of --check

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

50

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

51 def reverse_complement(sequence):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

52 complement = {

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

53 'A': 'T',

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

54 'C': 'G',

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

55 'G': 'C',

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

56 'T': 'A',

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

57 'N': 'N',

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

58 'M': 'K',

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

59 'R': 'Y',

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

60 'W': 'W',

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

61 'S': 'S',

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

62 'Y': 'R',

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

63 'K': 'M',

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

64 'V': 'B',

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

65 'H': 'D',

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

66 'D': 'H',

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

67 'B': 'V'

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

68 }

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

69 return "".join(complement[base] for base in reversed(sequence))

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

70

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

71 def mlst(assembly):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

72 subprocess.check_call("mlst -q --json mlst.json --scheme senterica_achtman_2 "+assembly+" >> data_log.txt 2>&1",shell=True)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

73 f = open("mlst.json",'r')

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

74 mlst_result = json.load(f)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

75 f.close()

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

76 st = mlst_result[0]['sequence_type']

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

77 alleles = mlst_result[0]['alleles']

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

78 return(st,alleles)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

79

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

80 def stringmlst(f1,f2,t,d):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

81 if t in ['1','2']:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

82 subprocess.check_call("stringMLST.py --predict -P "+d+"/kmer/salmonella -1 "+f1+" -2 "+f2+" -a /dev/stdout -o stringMLST.txt >> data_log.txt 2>&1",shell=True)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

83 elif t=='3':

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

84 subprocess.check_call("stringMLST.py --predict -P "+d+"/kmer/salmonella -1 "+f1+" -s -a /dev/stdout -o stringMLST.txt >> data_log.txt 2>&1",shell=True)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

85 f = "stringMLST.txt"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

86 mlst_result = open(f).readlines()[1].strip().split('\t')

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

87 st = mlst_result[-1]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

88 if st == '0':

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

89 st = '-'

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

90 k = ['aroC','dnaN','hemD','hisD','purE','sucA','thrA']

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

91 v = [mlst_result[1],mlst_result[2],mlst_result[3],mlst_result[4],mlst_result[5],mlst_result[6],mlst_result[7]]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

92 alleles = dict(zip(k,v))

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

93 return(st,alleles)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

94

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

95 def createKmerDict_reads(list_of_strings, kmer):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

96 kmer_table = {}

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

97 for string in list_of_strings:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

98 sequence = string.strip('\n')

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

99 for i in range(len(sequence) - kmer + 1):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

100 new_mer = sequence[i:i + kmer].upper()

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

101 new_mer_rc = reverse_complement(new_mer)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

102 if new_mer in kmer_table:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

103 kmer_table[new_mer.upper()] += 1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

104 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

105 kmer_table[new_mer.upper()] = 1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

106 if new_mer_rc in kmer_table:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

107 kmer_table[new_mer_rc.upper()] += 1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

108 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

109 kmer_table[new_mer_rc.upper()] = 1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

110 return kmer_table

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

111

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

112

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

113 def multifasta_dict(multifasta):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

114 multifasta_list = [

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

115 line.strip() for line in open(multifasta, 'r') if len(line.strip()) > 0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

116 ]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

117 headers = [i for i in multifasta_list if i[0] == '>']

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

118 multifasta_dict = {}

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

119 for h in headers:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

120 start = multifasta_list.index(h)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

121 for element in multifasta_list[start + 1:]:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

122 if element[0] == '>':

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

123 break

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

124 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

125 if h[1:] in multifasta_dict:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

126 multifasta_dict[h[1:]] += element

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

127 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

128 multifasta_dict[h[1:]] = element

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

129 return multifasta_dict

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

130

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

131

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

132 def multifasta_single_string(multifasta):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

133 multifasta_list = [

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

134 line.strip() for line in open(multifasta, 'r')

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

135 if (len(line.strip()) > 0) and (line.strip()[0] != '>')

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

136 ]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

137 return ''.join(multifasta_list)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

138

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

139

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

140 def chunk_a_long_sequence(long_sequence, chunk_size=60):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

141 chunk_list = []

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

142 steps = len(long_sequence) // 60 #how many chunks

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

143 for i in range(steps):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

144 chunk_list.append(long_sequence[i * chunk_size:(i + 1) * chunk_size])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

145 chunk_list.append(long_sequence[steps * chunk_size:len(long_sequence)])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

146 return chunk_list

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

147

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

148

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

149 def target_multifasta_kmerizer(multifasta, k, kmerDict):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

150 forward_length = 300 #if find the target, put forward 300 bases

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

151 reverse_length = 2200 #if find the target, put backward 2200 bases

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

152 chunk_size = 60 #it will firstly chunk the single long sequence to multiple smaller sequences, it controls the size of those smaller sequences

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

153 target_mers = []

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

154 long_single_string = multifasta_single_string(multifasta)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

155 multifasta_list = chunk_a_long_sequence(long_single_string, chunk_size)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

156 unit_length = len(multifasta_list[0])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

157 forward_lines = int(forward_length / unit_length) + 1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

158 reverse_lines = int(forward_length / unit_length) + 1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

159 start_num = 0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

160 end_num = 0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

161 for i in range(len(multifasta_list)):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

162 if i not in range(start_num, end_num): #avoid computational repetition

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

163 line = multifasta_list[i]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

164 start = int((len(line) - k) // 2)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

165 s1 = line[start:k + start]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

166 if s1 in kmerDict: #detect it is a potential read or not (use the middle part)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

167 if i - forward_lines >= 0:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

168 start_num = i - forward_lines

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

169 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

170 start_num = 0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

171 if i + reverse_lines <= len(multifasta_list) - 1:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

172 end_num = i + reverse_lines

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

173 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

174 end_num = len(multifasta_list) - 1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

175 target_list = [

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

176 x.strip() for x in multifasta_list[start_num:end_num]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

177 ]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

178 target_line = "".join(target_list)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

179 target_mers += [

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

180 k1 for k1 in createKmerDict_reads([str(target_line)], k)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

181 ] ##changed k to k1, just want to avoid the mixes of this "k" (kmer) to the "k" above (kmer length)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

182 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

183 pass

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

184 return set(target_mers)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

185

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

186

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

187 def target_read_kmerizer(file, k, kmerDict):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

188 i = 1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

189 n_reads = 0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

190 total_coverage = 0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

191 target_mers = []

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

192 if file.endswith(".gz"):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

193 file_content = io.BufferedReader(gzip.open(file))

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

194 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

195 file_content = open(file, "r").readlines()

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

196 for line in file_content:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

197 start = int((len(line) - k) // 2)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

198 if i % 4 == 2:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

199 if file.endswith(".gz"):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

200 s1 = line[start:k + start].decode()

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

201 line = line.decode()

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

202 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

203 s1 = line[start:k + start]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

204 if s1 in kmerDict: #detect it is a potential read or not (use the middle part)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

205 n_reads += 1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

206 total_coverage += len(line)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

207 target_mers += [

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

208 k1 for k1 in createKmerDict_reads([str(line)], k)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

209 ] #changed k to k1, just want to avoid the mixes of this "k" (kmer) to the "k" above (kmer length)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

210 i += 1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

211 if total_coverage >= 4000000:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

212 break

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

213 return set(target_mers)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

214

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

215

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

216 def minion_fasta_kmerizer(file, k, kmerDict):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

217 i = 1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

218 n_reads = 0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

219 total_coverage = 0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

220 target_mers = {}

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

221 for line in open(file):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

222 if i % 2 == 0:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

223 for kmer, rc_kmer in kmers(line.strip().upper(), k):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

224 if (kmer in kmerDict) or (rc_kmer in kmerDict):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

225 if kmer in target_mers:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

226 target_mers[kmer] += 1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

227 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

228 target_mers[kmer] = 1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

229 if rc_kmer in target_mers:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

230 target_mers[rc_kmer] += 1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

231 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

232 target_mers[rc_kmer] = 1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

233 i += 1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

234 return set([h for h in target_mers])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

235

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

236

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

237 def minion_fastq_kmerizer(file, k, kmerDict):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

238 i = 1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

239 n_reads = 0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

240 total_coverage = 0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

241 target_mers = {}

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

242 for line in open(file):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

243 if i % 4 == 2:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

244 for kmer, rc_kmer in kmers(line.strip().upper(), k):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

245 if (kmer in kmerDict) or (rc_kmer in kmerDict):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

246 if kmer in target_mers:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

247 target_mers[kmer] += 1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

248 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

249 target_mers[kmer] = 1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

250 if rc_kmer in target_mers:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

251 target_mers[rc_kmer] += 1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

252 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

253 target_mers[rc_kmer] = 1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

254 i += 1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

255 return set([h for h in target_mers])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

256

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

257

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

258 def multifasta_single_string2(multifasta):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

259 single_string = ''

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

260 with open(multifasta, 'r') as f:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

261 for line in f:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

262 if line.strip()[0] == '>':

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

263 pass

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

264 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

265 single_string += line.strip()

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

266 return single_string

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

267

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

268

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

269 def kmers(seq, k):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

270 rev_comp = reverse_complement(seq)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

271 for start in range(1, len(seq) - k + 1):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

272 yield seq[start:start + k], rev_comp[-(start + k):-start]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

273

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

274

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

275 def multifasta_to_kmers_dict(multifasta,k_size):#used to create database kmer set

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

276 multi_seq_dict = multifasta_dict(multifasta)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

277 lib_dict = {}

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

278 for h in multi_seq_dict:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

279 lib_dict[h] = set(

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

280 [k for k in createKmerDict_reads([multi_seq_dict[h]], k_size)])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

281 return lib_dict

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

282

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

283

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

284 def Combine(b, c):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

285 fliC_combinations = []

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

286 fliC_combinations.append(",".join(c))

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

287 temp_combinations = []

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

288 for i in range(len(b)):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

289 for x in itertools.combinations(b, i + 1):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

290 temp_combinations.append(",".join(x))

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

291 for x in temp_combinations:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

292 temp = []

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

293 for y in c:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

294 temp.append(y)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

295 temp.append(x)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

296 temp = ",".join(temp)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

297 temp = temp.split(",")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

298 temp.sort()

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

299 temp = ",".join(temp)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

300 fliC_combinations.append(temp)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

301 return fliC_combinations

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

302

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

303

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

304 def seqsero_from_formula_to_serotypes(Otype, fliC, fljB, special_gene_list,subspecies,ss):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

305 #like test_output_06012017.txt

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

306 #can add more varialbles like sdf-type, sub-species-type in future (we can conclude it into a special-gene-list)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

307 #from Initial_Conditions import phase1,phase2,phaseO,sero,subs,remove_list,rename_dict

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

308 if ss == 'ss2':

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

309 from Initial_Conditions_SS2 import phase1,phase2,phaseO,sero,subs,remove_list,rename_dict

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

310 elif ss == 'ss2s':

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

311 from Initial_Conditions_SS2S import phase1,phase2,phaseO,sero,subs,remove_list,rename_dict

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

312 rename_dict_not_anymore=[rename_dict[x] for x in rename_dict]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

313 rename_dict_all=rename_dict_not_anymore+list(rename_dict) #used for decide whether to

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

314 seronames = []

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

315 seronames_none_subspecies=[]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

316 for i in range(len(phase1)):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

317 fliC_combine = []

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

318 fljB_combine = []

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

319 if phaseO[i] == Otype: # no VII in KW, but it's there

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

320 ### for fliC, detect every possible combinations to avoid the effect of "["

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

321 if phase1[i].count("[") == 0:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

322 fliC_combine.append(phase1[i])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

323 elif phase1[i].count("[") >= 1:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

324 c = []

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

325 b = []

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

326 if phase1[i][0] == "[" and phase1[i][-1] == "]" and phase1[i].count(

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

327 "[") == 1:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

328 content = phase1[i].replace("[", "").replace("]", "")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

329 fliC_combine.append(content)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

330 fliC_combine.append("-")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

331 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

332 for x in phase1[i].split(","):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

333 if "[" in x:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

334 b.append(x.replace("[", "").replace("]", ""))

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

335 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

336 c.append(x)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

337 fliC_combine = Combine(

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

338 b, c

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

339 ) #Combine will offer every possible combinations of the formula, like f,[g],t: f,t f,g,t

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

340 ### end of fliC "[" detect

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

341 ### for fljB, detect every possible combinations to avoid the effect of "["

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

342 if phase2[i].count("[") == 0:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

343 fljB_combine.append(phase2[i])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

344 elif phase2[i].count("[") >= 1:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

345 d = []

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

346 e = []

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

347 if phase2[i][0] == "[" and phase2[i][-1] == "]" and phase2[i].count(

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

348 "[") == 1:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

349 content = phase2[i].replace("[", "").replace("]", "")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

350 fljB_combine.append(content)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

351 fljB_combine.append("-")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

352 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

353 for x in phase2[i].split(","):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

354 if "[" in x:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

355 d.append(x.replace("[", "").replace("]", ""))

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

356 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

357 e.append(x)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

358 fljB_combine = Combine(d, e)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

359 ### end of fljB "[" detect

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

360 new_fliC = fliC.split(

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

361 ","

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

362 ) #because some antigen like r,[i] not follow alphabetical order, so use this one to judge and can avoid missings

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

363 new_fliC.sort()

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

364 new_fliC = ",".join(new_fliC)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

365 new_fljB = fljB.split(",")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

366 new_fljB.sort()

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

367 new_fljB = ",".join(new_fljB)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

368 if (new_fliC in fliC_combine

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

369 or fliC in fliC_combine) and (new_fljB in fljB_combine

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

370 or fljB in fljB_combine):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

371 ######start, remove_list,rename_dict, added on 11/11/2018

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

372 if sero[i] not in remove_list:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

373 temp_sero=sero[i]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

374 if temp_sero in rename_dict:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

375 temp_sero=rename_dict[temp_sero] #rename if in the rename list

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

376 if temp_sero not in seronames:#the new sero may already included, if yes, then not consider

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

377 if subs[i] == subspecies:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

378 seronames.append(temp_sero)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

379 seronames_none_subspecies.append(temp_sero)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

380 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

381 pass

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

382 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

383 pass

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

384 ######end, added on 11/11/2018

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

385 #analyze seronames

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

386 subspecies_pointer=""

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

387 if len(seronames) == 0 and len(seronames_none_subspecies)!=0:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

388 ## ed_SL_06062020: for the subspecies mismatch between KW and SalmID

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

389 seronames=seronames_none_subspecies

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

390 #seronames=["N/A"]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

391 subspecies_pointer="1"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

392 #subspecies_pointer="0"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

393 ##

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

394 if len(seronames) == 0:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

395 seronames = [

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

396 "N/A (The predicted antigenic profile does not exist in the White-Kauffmann-Le Minor scheme)"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

397 ]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

398 star = ""

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

399 star_line = ""

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

400 if len(seronames) > 1: #there are two possible predictions for serotypes

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

401 star = "*"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

402 #changed 04072019

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

403 #star_line = "The predicted serotypes share the same general formula:\t" + Otype + ":" + fliC + ":" + fljB + "\n"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

404 if subspecies_pointer=="1" and len(seronames_none_subspecies)!=0:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

405 star="*"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

406 star_line = "This antigenic profile has been associated with serotype '"+(" or ").join(seronames)+"' in the Kauffman-White scheme. The existence of the same antigenic formula in multiple species or subspecies is well documented in the Kauffman-White Scheme. " + star_line ## ed_SL_03202021: changed for new output format

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

407 #star_line="The predicted O and H antigens correspond to serotype '"+(" or ").join(seronames)+"' in the Kauffmann-White scheme. The predicted subspecies by SalmID (github.com/hcdenbakker/SalmID) may not be consistent with subspecies designation in the Kauffmann-White scheme. " + star_line

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

408 #star_line="The formula with this subspieces prediction can't get a serotype in KW manual, and the serotyping prediction was made without considering it."+star_line

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

409 seronames=["N/A"] ## ed_SL_06062020

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

410 if Otype=="":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

411 Otype="-"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

412 predict_form = Otype + ":" + fliC + ":" + fljB

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

413 predict_sero = (" or ").join(seronames)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

414 ###special test for Enteritidis

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

415 if predict_form == "9:g,m:-":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

416 sdf = "-"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

417 for x in special_gene_list:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

418 if x.startswith("sdf"):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

419 sdf = "+"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

420 #star_line="Detected sdf gene, a marker to differentiate Gallinarum and Enteritidis"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

421 #star_line="sdf gene detected. "

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

422 star_line = "Detected Sdf I that is characteristic of commonly circulating strains of serotype Enteritidis. "

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

423 #predict_form = predict_form + " Sdf prediction:" + sdf

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

424 predict_form = predict_form #changed 04072019

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

425 if sdf == "-":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

426 star = "*"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

427 #star_line="Didn't detected sdf gene, a marker to differentiate Gallinarum and Enteritidis"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

428 #star_line="sdf gene not detected. "

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

429 star_line = "Sdf I that is characteristic of commonly circulating strains of serotype Enteritidis was not detected. "

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

430 #changed in 04072019, for new output

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

431 #star_line = "Additional characterization is necessary to assign a serotype to this strain. Commonly circulating strains of serotype Enteritidis are sdf+, although sdf- strains of serotype Enteritidis are known to exist. Serotype Gallinarum is typically sdf- but should be quite rare. Sdf- strains of serotype Enteritidis and serotype Gallinarum can be differentiated by phenotypic profile or genetic criteria.\n"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

432 #predict_sero = "Gallinarum/Enteritidis" #04132019, for new output requirement

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

433 predict_sero = "Gallinarum or Enteritidis"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

434 ###end of special test for Enteritidis

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

435 elif predict_form == "4:i:-":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

436 predict_sero = "I 4,[5],12:i:-" # change serotype name

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

437 elif predict_form == "4:r:-":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

438 predict_sero = "N/A (4:r:-)"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

439 elif predict_form == "4:b:-":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

440 predict_sero = "N/A (4:b:-)"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

441 #elif predict_form == "8:e,h:1,2": #removed after official merge of newport and bardo

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

442 #predict_sero = "Newport"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

443 #star = "*"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

444 #star_line = "Serotype Bardo shares the same antigenic profile with Newport, but Bardo is exceedingly rare."

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

445 claim = "The serotype(s) is/are the only serotype(s) with the indicated antigenic profile currently recognized in the Kauffmann White Scheme. New serotypes can emerge and the possibility exists that this antigenic profile may emerge in a different subspecies. Identification of strains to the subspecies level should accompany serotype determination; the same antigenic profile in different subspecies is considered different serotypes.\n"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

446 if "N/A" in predict_sero:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

447 claim = ""

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

448 #special test for Typhimurium

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

449 if "Typhimurium" in predict_sero or predict_form == "4:i:-":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

450 normal = 0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

451 mutation = 0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

452 for x in special_gene_list:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

453 if "oafA-O-4_full" in x:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

454 normal = float(special_gene_list[x])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

455 elif "oafA-O-4_5-" in x:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

456 mutation = float(special_gene_list[x])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

457 if normal > mutation:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

458 pass

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

459 elif normal < mutation:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

460 #predict_sero = predict_sero.strip() + "(O5-)"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

461 predict_sero = predict_sero.strip() #diable special sero for new output requirement, 04132019

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

462 star = "*"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

463 #star_line = "Detected the deletion of O5-."

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

464 star_line = "Detected a deletion in gene oafA that causes O5- variant of Typhimurium. "

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

465 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

466 pass

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

467 #special test for Paratyphi B

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

468 if "Paratyphi B" in predict_sero or predict_form == "4:b:-":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

469 normal = 0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

470 mutation = 0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

471 for x in special_gene_list:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

472 if "gntR-family-regulatory-protein_dt-positive" in x:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

473 normal = float(special_gene_list[x])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

474 elif "gntR-family-regulatory-protein_dt-negative" in x:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

475 mutation = float(special_gene_list[x])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

476 #print(normal,mutation)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

477 if normal > mutation:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

478 #predict_sero = predict_sero.strip() + "(dt+)" #diable special sero for new output requirement, 04132019

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

479 predict_sero = predict_sero.strip()+' var. L(+) tartrate+' if "Paratyphi B" in predict_sero else predict_sero.strip()

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

480 star = "*"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

481 #star_line = "Didn't detect the SNP for dt- which means this isolate is a Paratyphi B variant L(+) tartrate(+)."

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

482 star_line = "The SNP in gene STM3356 that is associated with the d-Tartrate nonfermenting phenotype characteristic of the typhoidal pathotype was not detected. "

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

483 elif normal < mutation:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

484 #predict_sero = predict_sero.strip() + "(dt-)" #diable special sero for new output requirement, 04132019

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

485 predict_sero = predict_sero.strip()

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

486 star = "*"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

487 #star_line = "Detected the SNP for d-Tartrate nonfermenting phenotype of Paratyphi B. "

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

488 star_line = "Detected the SNP in gene STM3356 that is associated with the d-Tartrate nonfermenting phenotype characteristic of the typhoidal pathotype. "

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

489 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

490 star = "*"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

491 #star_line = " Failed to detect the SNP for dt-, can't decide it's a Paratyphi B variant L(+) tartrate(+) or not."

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

492 star_line = " " ## ed_SL_05152019: do not report this situation.

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

493 #special test for O13,22 and O13,23

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

494 ### add comment for any O2 call. 06052024

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

495 if Otype=='2' and predict_sero not in ['Nitra','Kiel','Koessen']:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

496 star_line = 'O2 is typically a O9 rfb serotype with a mutation that results in a different sugar being placed in O antigen. SS2S detects only one group O2 serotype, Paratyphi A. This genome may be a variant of a group O9 serotype.'

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

497 ###

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

498 if Otype=="13":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

499 #ex_dir = os.path.dirname(os.path.realpath(__file__))

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

500 ex_dir = os.path.abspath(os.path.join(os.path.dirname(os.path.dirname(__file__)),'seqsero2s_db')) # ed_SL_09152019

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

501 f = open(ex_dir + '/special.pickle', 'rb')

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

502 special = pickle.load(f)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

503 O22_O23=special['O22_O23']

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

504 if predict_sero.split(" or ")[0] in O22_O23[-1] and predict_sero.split(" or ")[0] not in rename_dict_all:#if in rename_dict_all, then it means already merged, no need to analyze

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

505 #if predict_sero.split(" or ")[0] in O22_O23[-1]: # Report O22 vs O23 result for O13 serotypes. 12232024

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

506 O22_score=0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

507 O23_score=0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

508 for x in special_gene_list:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

509 if "O:22" in x:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

510 O22_score = O22_score+float(special_gene_list[x])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

511 elif "O:23" in x:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

512 O23_score = O23_score+float(special_gene_list[x])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

513 #print(O22_score,O23_score)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

514 for z in O22_O23[0]:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

515 if predict_sero.split(" or ")[0] in z:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

516 if O22_score > O23_score:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

517 star = "*"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

518 star_line = "Detected a genetic marker (galE allele) for ancillary O22."

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

519 #star_line = "Detected O22 specific genes to further differenciate '"+predict_sero+"'." #diabled for new output requirement, 04132019

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

520 predict_sero = z[0]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

521 elif O22_score < O23_score:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

522 star = "*"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

523 star_line = "Detected a genetic marker (galE allele) for ancillary O23."

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

524 #star_line = "Detected O23 specific genes to further differenciate '"+predict_sero+"'." #diabled for new output requirement, 04132019

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

525 predict_sero = z[1]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

526 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

527 star = "*"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

528 star_line = "Fail to detect genetic markers (galE alleles) for ancillary O22 and O23."

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

529 #star_line = "Fail to detect O22/O23 specific genes." #diabled for new output requirement, 04132019

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

530 if " or " in predict_sero:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

531 star_line = star_line + "The predicted serotypes share the same general formula: " + Otype + ":" + fliC + ":" + fljB + " and can be differentiated by additional analysis. "

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

532 #special test for O6,8

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

533 #merge_O68_list=["Blockley","Bovismorbificans","Hadar","Litchfield","Manhattan","Muenchen"] #remove 11/11/2018, because already in merge list

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

534 #for x in merge_O68_list:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

535 # if x in predict_sero:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

536 # predict_sero=x

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

537 # star=""

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

538 # star_line=""

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

539 #special test for Montevideo; most of them are monophasic

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

540 #if "Montevideo" in predict_sero and "1,2,7" in predict_form: #remove 11/11/2018, because already in merge list

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

541 #star="*"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

542 #star_line="Montevideo is almost always monophasic, having an antigen called for the fljB position may be a result of Salmonella-Salmonella contamination."

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

543 return predict_form, predict_sero, star, star_line, claim

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

544 ### End of SeqSero Kmer part

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

545

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

546 ### Begin of SeqSero2 allele prediction and output

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

547 def xml_parse_score_comparision_seqsero(xmlfile):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

548 #used to do seqsero xml analysis

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

549 from Bio.Blast import NCBIXML

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

550 handle=open(xmlfile)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

551 handle=NCBIXML.parse(handle)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

552 handle=list(handle)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

553 List=[]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

554 List_score=[]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

555 List_ids=[]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

556 List_query_region=[]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

557 for i in range(len(handle)):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

558 if len(handle[i].alignments)>0:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

559 for j in range(len(handle[i].alignments)):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

560 score=0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

561 ids=0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

562 cover_region=set() #fixed problem that repeated calculation leading percentage > 1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

563 List.append(handle[i].query.strip()+"___"+handle[i].alignments[j].hit_def)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

564 for z in range(len(handle[i].alignments[j].hsps)):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

565 hsp=handle[i].alignments[j].hsps[z]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

566 temp=set(range(hsp.query_start,hsp.query_end))

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

567 if len(cover_region)==0:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

568 cover_region=cover_region|temp

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

569 fraction=1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

570 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

571 fraction=1-len(cover_region&temp)/float(len(temp))

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

572 cover_region=cover_region|temp

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

573 if "last" in handle[i].query or "first" in handle[i].query:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

574 score+=hsp.bits*fraction

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

575 ids+=float(hsp.identities)/handle[i].query_length*fraction

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

576 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

577 score+=hsp.bits*fraction

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

578 ids+=float(hsp.identities)/handle[i].query_length*fraction

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

579 List_score.append(score)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

580 List_ids.append(ids)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

581 List_query_region.append(cover_region)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

582 temp=zip(List,List_score,List_ids,List_query_region)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

583 Final_list=sorted(temp, key=lambda d:d[1], reverse = True)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

584 return Final_list

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

585

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

586

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

587 def Uniq(L,sort_on_fre="none"): #return the uniq list and the count number

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

588 Old=L

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

589 L.sort()

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

590 L = [L[i] for i in range(len(L)) if L[i] not in L[:i]]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

591 count=[]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

592 for j in range(len(L)):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

593 y=0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

594 for x in Old:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

595 if L[j]==x:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

596 y+=1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

597 count.append(y)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

598 if sort_on_fre!="none":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

599 d=zip(*sorted(zip(count, L)))

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

600 L=d[1]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

601 count=d[0]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

602 return (L,count)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

603

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

604 def judge_fliC_or_fljB_from_head_tail_for_one_contig(nodes_vs_score_list):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

605 #used to predict it's fliC or fljB for one contig, based on tail and head score, but output the score difference,if it is very small, then not reliable, use blast score for whole contig to test

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

606 #this is mainly used for

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

607 a=nodes_vs_score_list

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

608 fliC_score=0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

609 fljB_score=0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

610 for z in a:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

611 if "fliC" in z[0]:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

612 fliC_score+=z[1]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

613 elif "fljB" in z[0]:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

614 fljB_score+=z[1]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

615 if fliC_score>=fljB_score:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

616 role="fliC"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

617 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

618 role="fljB"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

619 return (role,abs(fliC_score-fljB_score))

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

620

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

621 def judge_fliC_or_fljB_from_whole_contig_blast_score_ranking(node_name,Final_list,Final_list_passed):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

622 #used to predict contig is fliC or fljB, if the differnce score value on above head_and_tail is less than 10 (quite small)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

623 #also used when no head or tail got blasted score for the contig

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

624 role=""

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

625 for z in Final_list_passed:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

626 if node_name in z[0]:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

627 role=z[0].split("_")[0]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

628 break

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

629 return role

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

630

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

631 def fliC_or_fljB_judge_from_head_tail_sequence(nodes_list,tail_head_list,Final_list,Final_list_passed):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

632 #nodes_list is the c created by c,d=Uniq(nodes) in below function

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

633 first_target=""

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

634 role_list=[]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

635 for x in nodes_list:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

636 a=[]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

637 role=""

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

638 for y in tail_head_list:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

639 if x in y[0]:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

640 a.append(y)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

641 if len(a)==4:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

642 role,diff=judge_fliC_or_fljB_from_head_tail_for_one_contig(a)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

643 if diff<20:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

644 role=judge_fliC_or_fljB_from_whole_contig_blast_score_ranking(x,Final_list,Final_list_passed)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

645 elif len(a)==3:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

646 ###however, if the one with highest score is the fewer one, compare their accumulation score

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

647 role,diff=judge_fliC_or_fljB_from_head_tail_for_one_contig(a)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

648 if diff<20:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

649 role=judge_fliC_or_fljB_from_whole_contig_blast_score_ranking(x,Final_list,Final_list_passed)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

650 ###end of above score comparison

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

651 elif len(a)==2:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

652 #must on same node, if not, then decide with unit blast score, blast-score/length_of_special_sequence(30 or 37)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

653 temp=[]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

654 for z in a:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

655 temp.append(z[0].split("_")[0])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

656 m,n=Uniq(temp)#should only have one choice, but weird situation might occur too

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

657 if len(m)==1:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

658 pass

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

659 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

660 pass

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

661 role,diff=judge_fliC_or_fljB_from_head_tail_for_one_contig(a)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

662 if diff<20:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

663 role=judge_fliC_or_fljB_from_whole_contig_blast_score_ranking(x,Final_list,Final_list_passed)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

664 ###need to desgin a algorithm to guess most possible situation for nodes_list, See the situations of test evaluation

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

665 elif len(a)==1:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

666 #that one

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

667 role,diff=judge_fliC_or_fljB_from_head_tail_for_one_contig(a)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

668 if diff<20:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

669 role=judge_fliC_or_fljB_from_whole_contig_blast_score_ranking(x,Final_list,Final_list_passed)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

670 #need to evaluate, in future, may set up a cut-off, if not met, then just find Final_list_passed best match,like when "a==0"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

671 else:#a==0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

672 #use Final_list_passed best match

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

673 for z in Final_list_passed:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

674 if x in z[0]:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

675 role=z[0].split("_")[0]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

676 break

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

677 #print x,role,len(a)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

678 role_list.append((role,x))

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

679 if len(role_list)==2:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

680 if role_list[0][0]==role_list[1][0]:#this is the most cocmmon error, two antigen were assigned to same phase

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

681 #just use score to do a final test

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

682 role_list=[]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

683 for x in nodes_list:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

684 role=judge_fliC_or_fljB_from_whole_contig_blast_score_ranking(x,Final_list,Final_list_passed)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

685 role_list.append((role,x))

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

686 return role_list

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

687

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

688 def decide_contig_roles_for_H_antigen(Final_list,Final_list_passed):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

689 #used to decide which contig is FliC and which one is fljB

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

690 contigs=[]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

691 nodes=[]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

692 for x in Final_list_passed:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

693 if x[0].startswith("fl") and "last" not in x[0] and "first" not in x[0]:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

694 nodes.append(x[0].split("___")[1].strip())

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

695 c,d=Uniq(nodes)#c is node_list

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

696 #print c

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

697 tail_head_list=[x for x in Final_list if ("last" in x[0] or "first" in x[0])]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

698 roles=fliC_or_fljB_judge_from_head_tail_sequence(c,tail_head_list,Final_list,Final_list_passed)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

699 return roles

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

700

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

701 def decide_O_type_and_get_special_genes(Final_list,Final_list_passed):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

702 #decide O based on Final_list

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

703 O_choice="?"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

704 O_list=[]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

705 special_genes={}

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

706 nodes=[]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

707 for x in Final_list_passed:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

708 if x[0].startswith("O-"):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

709 nodes.append(x[0].split("___")[1].strip())

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

710 elif not x[0].startswith("fl"):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

711 special_genes[x[0]]=x[2]#08172018, x[2] changed from x[-1]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

712 ##print("special_genes:",special_genes)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

713 c,d=Uniq(nodes)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

714 #print "potential O antigen contig",c

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

715 final_O=[]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

716 O_nodes_list=[]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

717 for x in c:#c is the list for contigs

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

718 temp=0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

719 for y in Final_list_passed:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

720 if x in y[0] and y[0].startswith("O-"):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

721 final_O.append(y)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

722 break

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

723 ### O contig has the problem of two genes on same contig, so do additional test

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

724 potenial_new_gene=""

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

725 for x in final_O:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

726 pointer=0 #for genes merged or not

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

727 #not consider O-1,3,19_not_in_3,10, too short compared with others

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

728 if "O-1,3,19_not_in_3,10" not in x[0] and int(x[0].split("__")[1].split("___")[0])*x[2]+850 <= int(x[0].split("length_")[1].split("_")[0]):#gene length << contig length; for now give 300*2 (for secureity can use 400*2) as flank region

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

729 pointer=x[0].split("___")[1].strip()#store the contig name

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

730 print(pointer)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

731 if pointer!=0:#it has potential merge event

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

732 for y in Final_list:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

733 if pointer in y[0] and y not in final_O and (y[1]>=int(y[0].split("__")[1].split("___")[0])*1.5 or (y[1]>=int(y[0].split("__")[1].split("___")[0])*y[2] and y[1]>=400)):#that's a realtively strict filter now; if passed, it has merge event and add one more to final_O

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

734 potenial_new_gene=y

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

735 #print(potenial_new_gene)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

736 break

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

737 if potenial_new_gene!="":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

738 print("two differnt genes in same contig, fix it for O antigen")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

739 print(potenial_new_gene[:3])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

740 pointer=0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

741 for y in final_O:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

742 if y[0].split("___")[-1]==potenial_new_gene[0].split("___")[-1]:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

743 pointer=1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

744 if pointer!=0: #changed to consider two genes in same contig

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

745 final_O.append(potenial_new_gene)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

746 ### end of the two genes on same contig test

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

747 final_O=sorted(final_O,key=lambda x: x[2], reverse=True)#sorted

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

748 if len(final_O)==0 or (len(final_O)==1 and "O-1,3,19_not_in_3,10" in final_O[0][0]):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

749 #print "$$$No Otype, due to no hit"#may need to be changed

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

750 O_choice="-"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

751 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

752 highest_O_coverage=max([float(x[0].split("_cov_")[-1].split("_")[0]) for x in final_O if "O-1,3,19_not_in_3,10" not in x[0]])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

753 O_list=[]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

754 O_list_less_contamination=[]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

755 for x in final_O:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

756 if not "O-1,3,19_not_in_3,10__130" in x[0]:#O-1,3,19_not_in_3,10 is too small, which may affect further analysis; to avoid contamination affect, use 0.15 of highest coverage as cut-off

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

757 O_list.append(x[0].split("__")[0])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

758 O_nodes_list.append(x[0].split("___")[1])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

759 if float(x[0].split("_cov_")[-1].split("_")[0])>highest_O_coverage*0.15:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

760 O_list_less_contamination.append(x[0].split("__")[0])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

761 ### special test for O9,46 and O3,10 family

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

762 if ("O-9,46_wbaV" in O_list or "O-9,46_wbaV-from-II-9,12:z29:1,5-SRR1346254" in O_list) and O_list_less_contamination[0].startswith("O-9,"):#not sure should use and float(O9_wbaV)/float(num_1) > 0.1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

763 if "O-9,46_wzy" in O_list or "O-9,46_wzy_partial" in O_list:#and float(O946_wzy)/float(num_1) > 0.1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

764 O_choice="O-9,46"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

765 #print "$$$Most possilble Otype: O-9,46"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

766 elif "O-9,46,27_partial_wzy" in O_list:#and float(O94627)/float(num_1) > 0.1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

767 O_choice="O-9,46,27"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

768 #print "$$$Most possilble Otype: O-9,46,27"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

769 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

770 O_choice="O-9"#next, detect O9 vs O2?

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

771 O2=0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

772 O9=0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

773 for z in special_genes:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

774 if "tyr-O-9" in z and special_genes[z] > O9: ##20240322, add "special_genes[z] > O9" to avoid misidentification of O9 to O2 that caused by multiple tyr-O-9 contigs

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

775 O9=special_genes[z]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

776 elif "tyr-O-2" in z and special_genes[z] > O2: ##20240322, add "special_genes[z] > O2"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

777 O2=special_genes[z]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

778 if O2>O9:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

779 O_choice="O-2"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

780 elif O2<O9:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

781 pass

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

782 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

783 pass

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

784 #print "$$$No suitable one, because can't distinct it's O-9 or O-2, but O-9 has a more possibility."

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

785 elif ("O-3,10_wzx" in O_list) and ("O-9,46_wzy" in O_list) and (O_list[0].startswith("O-3,10") or O_list_less_contamination[0].startswith("O-9,46_wzy")):#and float(O310_wzx)/float(num_1) > 0.1 and float(O946_wzy)/float(num_1) > 0.1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

786 if "O-3,10_not_in_1,3,19" in O_list:#and float(O310_no_1319)/float(num_1) > 0.1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

787 O_choice="O-3,10"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

788 #print "$$$Most possilble Otype: O-3,10 (contain O-3,10_not_in_1,3,19)"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

789 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

790 O_choice="O-1,3,19"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

791 #print "$$$Most possilble Otype: O-1,3,19 (not contain O-3,10_not_in_1,3,19)"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

792 ### end of special test for O9,46 and O3,10 family

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

793 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

794 try:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

795 max_score=0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

796 for x in final_O:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

797 if x[2]>=max_score and float(x[0].split("_cov_")[-1].split("_")[0])>highest_O_coverage*0.15:#use x[2],08172018, the "coverage identity = cover_length * identity"; also meet coverage threshold

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

798 max_score=x[2]#change from x[-1] to x[2],08172018

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

799 O_choice=x[0].split("_")[0]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

800 if O_choice=="O-1,3,19":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

801 O_choice=final_O[1][0].split("_")[0]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

802 #print "$$$Most possilble Otype: ",O_choice

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

803 except:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

804 pass

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

805 #print "$$$No suitable Otype, or failure of mapping (please check the quality of raw reads)"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

806 if O_choice=="O-9,46,27" and len(O_list)==2 and "O-4_wzx" in O_list: #special for very low chance sitatuion between O4 and O9,27,46, this is for serotypes like Bredeney and Schwarzengrund (normallly O-4 will have higher score, but sometimes sequencing quality may affect the prediction)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

807 O_choice="O-4"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

808 #print "O:",O_choice,O_nodes_list

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

809 Otypes=[]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

810 for x in O_list:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

811 if x!="O-1,3,19_not_in_3,10":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

812 if "O-9,46_" not in x:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

813 Otypes.append(x.split("_")[0])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

814 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

815 Otypes.append(x.split("-from")[0])#O-9,46_wbaV-from-II-9,12:z29:1,5-SRR1346254

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

816 #Otypes=[x.split("_")[0] for x in O_list if x!="O-1,3,19_not_in_3,10"]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

817 Otypes_uniq,Otypes_fre=Uniq(Otypes)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

818 contamination_O=""

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

819 if O_choice=="O-9,46,27" or O_choice=="O-3,10" or O_choice=="O-1,3,19":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

820 if len(Otypes_uniq)>2:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

821 contamination_O="potential contamination from O antigen signals"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

822 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

823 if len(Otypes_uniq)>1:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

824 if O_choice=="O-4" and len(Otypes_uniq)==2 and "O-9,46,27" in Otypes_uniq: #for special 4,12,27 case such as Bredeney and Schwarzengrund

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

825 contamination_O=""

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

826 elif O_choice=="O-9,46" and len(Otypes_uniq)==2 and "O-9,46_wbaV" in Otypes_uniq and "O-9,46_wzy" in Otypes_uniq: #for special 4,12,27 case such as Bredeney and Schwarzengrund

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

827 contamination_O=""

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

828 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

829 contamination_O="potential contamination from O antigen signals"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

830 return O_choice,O_nodes_list,special_genes,final_O,contamination_O,Otypes_uniq

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

831 ### End of SeqSero2 allele prediction and output

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

832

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

833 def get_input_files(make_dir,input_file,data_type,dirpath):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

834 #tell input files from datatype

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

835 #"<int>: '1'(pair-end reads, interleaved),'2'(pair-end reads, seperated),'3'(single-end reads), '4'(assembly),'5'(nanopore fasta),'6'(nanopore fastq)"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

836 for_fq=""

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

837 rev_fq=""

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

838 os.chdir(make_dir)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

839 if data_type=="1":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

840 input_file=input_file[0].split("/")[-1]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

841 if input_file.endswith(".sra"):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

842 subprocess.check_call("fastq-dump --split-files "+input_file,shell=True)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

843 for_fq=input_file.replace(".sra","_1.fastq")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

844 rev_fq=input_file.replace(".sra","_2.fastq")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

845 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

846 core_id=input_file.split(".fastq")[0].split(".fq")[0]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

847 for_fq=core_id+"_1.fastq"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

848 rev_fq=core_id+"_2.fastq"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

849 if input_file.endswith(".gz"):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

850 subprocess.check_call("gzip -dc "+input_file+" | "+dirpath+"/deinterleave_fastq.sh "+for_fq+" "+rev_fq,shell=True)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

851 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

852 subprocess.check_call("cat "+input_file+" | "+dirpath+"/deinterleave_fastq.sh "+for_fq+" "+rev_fq,shell=True)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

853 elif data_type=="2":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

854 for_fq=input_file[0].split("/")[-1]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

855 rev_fq=input_file[1].split("/")[-1]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

856 elif data_type=="3":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

857 input_file=input_file[0].split("/")[-1]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

858 if input_file.endswith(".sra"):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

859 subprocess.check_call("fastq-dump --split-files "+input_file,shell=True)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

860 for_fq=input_file.replace(".sra","_1.fastq")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

861 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

862 for_fq=input_file

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

863 elif data_type in ["4","5","6"]:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

864 for_fq=input_file[0].split("/")[-1]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

865 os.chdir("..")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

866 return for_fq,rev_fq

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

867

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

868 def predict_O_and_H_types(Final_list,Final_list_passed,new_fasta):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

869 #get O and H types from Final_list from blast parsing; allele mode

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

870 from Bio import SeqIO

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

871 fliC_choice="-"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

872 fljB_choice="-"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

873 fliC_contig="NA"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

874 fljB_contig="NA"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

875 fliC_region=set([0])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

876 fljB_region=set([0,])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

877 fliC_length=0 #can be changed to coverage in future; in 03292019, changed to ailgned length

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

878 fljB_length=0 #can be changed to coverage in future; in 03292019, changed to ailgned length

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

879 O_choice="-"#no need to decide O contig for now, should be only one

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

880 O_choice,O_nodes,special_gene_list,O_nodes_roles,contamination_O,Otypes_uniq=decide_O_type_and_get_special_genes(Final_list,Final_list_passed)#decide the O antigen type and also return special-gene-list for further identification

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

881 O_choice=O_choice.split("-")[-1].strip()

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

882 if (O_choice=="1,3,19" and len(O_nodes_roles)==1 and "1,3,19" in O_nodes_roles[0][0]) or O_choice=="":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

883 O_choice="-"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

884 H_contig_roles=decide_contig_roles_for_H_antigen(Final_list,Final_list_passed)#decide the H antigen contig is fliC or fljB

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

885 #add alignment locations, used for further selection, 03312019

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

886 for i in range(len(H_contig_roles)):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

887 x=H_contig_roles[i]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

888 for y in Final_list_passed:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

889 if x[1] in y[0] and y[0].startswith(x[0]):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

890 H_contig_roles[i]+=H_contig_roles[i]+(y[-1],)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

891 break

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

892 log_file=open("SeqSero_log.txt","a")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

893 extract_file=open("Extracted_antigen_alleles.fasta","a")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

894 handle_fasta=list(SeqIO.parse(new_fasta,"fasta"))

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

895

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

896 #print("O_contigs:")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

897 log_file.write("O_contigs:\n")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

898 extract_file.write("#Sequences with antigen signals (if the micro-assembled contig only covers the flanking region, it will not be used for contamination analysis)\n")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

899 extract_file.write("#O_contigs:\n")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

900 for x in O_nodes_roles:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

901 if "O-1,3,19_not_in_3,10" not in x[0]:#O-1,3,19_not_in_3,10 is just a small size marker

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

902 #print(x[0].split("___")[-1],x[0].split("__")[0],"blast score:",x[1],"identity%:",str(round(x[2]*100,2))+"%",str(min(x[-1]))+" to "+str(max(x[-1])))

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

903 log_file.write(x[0].split("___")[-1]+" "+x[0].split("__")[0]+"; "+"blast score: "+str(x[1])+" identity%: "+str(round(x[2]*100,2))+"%; alignment from "+str(min(x[-1]))+" to "+str(max(x[-1]))+" of antigen\n")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

904 title=">"+x[0].split("___")[-1]+" "+x[0].split("__")[0]+"; "+"blast score: "+str(x[1])+" identity%: "+str(round(x[2]*100,2))+"%; alignment from "+str(min(x[-1]))+" to "+str(max(x[-1]))+" of antigen\n"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

905 seqs=""

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

906 for z in handle_fasta:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

907 if x[0].split("___")[-1]==z.description:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

908 seqs=str(z.seq)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

909 extract_file.write(title+seqs+"\n")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

910 if len(H_contig_roles)!=0:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

911 highest_H_coverage=max([float(x[1].split("_cov_")[-1].split("_")[0]) for x in H_contig_roles]) #less than highest*0.1 would be regarded as contamination and noises, they will still be considered in contamination detection and logs, but not used as final serotype output

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

912 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

913 highest_H_coverage=0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

914 for x in H_contig_roles:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

915 #if multiple choices, temporately select the one with longest length for now, will revise in further change

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

916 if "fliC" == x[0] and len(x[-1])>=fliC_length and x[1] not in O_nodes and float(x[1].split("_cov_")[-1].split("_")[0])>highest_H_coverage*0.13:#remember to avoid the effect of O-type contig, so should not in O_node list

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

917 fliC_contig=x[1]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

918 fliC_length=len(x[-1])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

919 elif "fljB" == x[0] and len(x[-1])>=fljB_length and x[1] not in O_nodes and float(x[1].split("_cov_")[-1].split("_")[0])>highest_H_coverage*0.13:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

920 fljB_contig=x[1]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

921 fljB_length=len(x[-1])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

922 for x in Final_list_passed:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

923 if fliC_choice=="-" and "fliC_" in x[0] and fliC_contig in x[0]:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

924 fliC_choice=x[0].split("_")[1]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

925 elif fljB_choice=="-" and "fljB_" in x[0] and fljB_contig in x[0]:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

926 fljB_choice=x[0].split("_")[1]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

927 elif fliC_choice!="-" and fljB_choice!="-":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

928 break

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

929 #now remove contigs not in middle core part

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

930 first_allele="NA"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

931 first_allele_percentage=0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

932 for x in Final_list:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

933 if x[0].startswith("fliC") or x[0].startswith("fljB"):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

934 first_allele=x[0].split("__")[0] #used to filter those un-middle contigs

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

935 first_allele_percentage=x[2]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

936 break

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

937 additional_contigs=[]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

938 for x in Final_list:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

939 if first_allele in x[0]:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

940 if (fliC_contig == x[0].split("___")[-1]):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

941 fliC_region=x[3]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

942 elif fljB_contig!="NA" and (fljB_contig == x[0].split("___")[-1]):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

943 fljB_region=x[3]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

944 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

945 if x[1]*1.1>int(x[0].split("___")[1].split("_")[3]):#loose threshold by multiplying 1.1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

946 additional_contigs.append(x)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

947 #else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

948 #print x[:3]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

949 #we can just use the fljB region (or fliC depends on size), no matter set() or contain a large locations (without middle part); however, if none of them is fully assembled, use 500 and 1200 as conservative cut-off

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

950 if first_allele_percentage>0.9:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

951 if len(fliC_region)>len(fljB_region) and (max(fljB_region)-min(fljB_region))>1000:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

952 target_region=fljB_region|(fliC_region-set(range(min(fljB_region),max(fljB_region)))) #fljB_region|(fliC_region-set(range(min(fljB_region),max(fljB_region))))

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

953 elif len(fliC_region)<len(fljB_region) and (max(fliC_region)-min(fliC_region))>1000:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

954 target_region=fliC_region|(fljB_region-set(range(min(fliC_region),max(fliC_region)))) #fljB_region|(fliC_region-set(range(min(fljB_region),max(fljB_region))))

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

955 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

956 target_region=set()#doesn't do anything

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

957 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

958 target_region=set()#doesn't do anything

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

959 #print(target_region)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

960 #print(additional_contigs)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

961 target_region2=set(list(range(0,525))+list(range(1200,1700)))#I found to use 500 to 1200 as special region would be best

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

962 target_region=target_region2|target_region

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

963 for x in additional_contigs:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

964 removal=0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

965 contig_length=int(x[0].split("___")[1].split("length_")[-1].split("_")[0])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

966 if fljB_contig not in x[0] and fliC_contig not in x[0] and len(target_region&x[3])/float(len(x[3]))>0.65 and contig_length*0.5<len(x[3])<contig_length*1.5: #consider length and alignment length for now, but very loose,0.5 and 1.5 as cut-off

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

967 removal=1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

968 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

969 if first_allele_percentage > 0.9 and float(x[0].split("__")[1].split("___")[0])*x[2]/len(x[-1])>0.96:#if high similiarity with middle part of first allele (first allele >0.9, already cover middle part)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

970 removal=1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

971 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

972 pass

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

973 if removal==1:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

974 for y in H_contig_roles:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

975 if y[1] in x[0]:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

976 H_contig_roles.remove(y)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

977 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

978 pass

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

979 #print(x[:3],contig_length,len(target_region&x[3])/float(len(x[3])),contig_length*0.5,len(x[3]),contig_length*1.5)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

980 #end of removing none-middle contigs

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

981 #print("H_contigs:")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

982 log_file.write("H_contigs:\n")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

983 extract_file.write("#H_contigs:\n")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

984 H_contig_stat=[]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

985 H1_cont_stat={}

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

986 H2_cont_stat={}

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

987 for i in range(len(H_contig_roles)):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

988 x=H_contig_roles[i]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

989 a=0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

990 for y in Final_list_passed:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

991 if x[1] in y[0] and y[0].startswith(x[0]):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

992 if "first" in y[0] or "last" in y[0]: #this is the final filter to decide it's fliC or fljB, if can't pass, then can't decide

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

993 for y in Final_list_passed: #it's impossible to has the "first" and "last" allele as prediction, so re-do it

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

994 if x[1] in y[0]:#it's very possible to be third phase allele, so no need to make it must be fliC or fljB

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

995 #print(x[1],"can't_decide_fliC_or_fljB",y[0].split("_")[1],"blast_score:",y[1],"identity%:",str(round(y[2]*100,2))+"%",str(min(y[-1]))+" to "+str(max(y[-1])))

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

996 log_file.write(x[1]+" "+x[0]+" "+y[0].split("_")[1]+"; "+"blast score: "+str(y[1])+" identity%: "+str(round(y[2]*100,2))+"%; alignment from "+str(min(y[-1]))+" to "+str(max(y[-1]))+" of antigen\n")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

997 H_contig_roles[i]="can't decide fliC or fljB, may be third phase"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

998 title=">"+x[1]+" "+x[0]+" "+y[0].split("_")[1]+"; "+"blast score: "+str(y[1])+" identity%: "+str(round(y[2]*100,2))+"%; alignment from "+str(min(y[-1]))+" to "+str(max(y[-1]))+" of antiten\n"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

999 seqs=""

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1000 for z in handle_fasta:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1001 if x[1]==z.description:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1002 seqs=str(z.seq)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1003 extract_file.write(title+seqs+"\n")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1004 break

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1005 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1006 #print(x[1],x[0],y[0].split("_")[1],"blast_score:",y[1],"identity%:",str(round(y[2]*100,2))+"%",str(min(y[-1]))+" to "+str(max(y[-1])))

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1007 log_file.write(x[1]+" "+x[0]+" "+y[0].split("_")[1]+"; "+"blast score: "+str(y[1])+" identity%: "+str(round(y[2]*100,2))+"%; alignment from "+str(min(y[-1]))+" to "+str(max(y[-1]))+" of antigen\n")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1008 title=">"+x[1]+" "+x[0]+" "+y[0].split("_")[1]+"; "+"blast score: "+str(y[1])+" identity%: "+str(round(y[2]*100,2))+"%; alignment from "+str(min(y[-1]))+" to "+str(max(y[-1]))+" of antigen\n"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1009 seqs=""

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1010 for z in handle_fasta:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1011 if x[1]==z.description:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1012 seqs=str(z.seq)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1013 extract_file.write(title+seqs+"\n")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1014 if x[0]=="fliC":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1015 if y[0].split("_")[1] not in H1_cont_stat:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1016 H1_cont_stat[y[0].split("_")[1]]=y[2]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1017 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1018 H1_cont_stat[y[0].split("_")[1]]+=y[2]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1019 if x[0]=="fljB":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1020 if y[0].split("_")[1] not in H2_cont_stat:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1021 H2_cont_stat[y[0].split("_")[1]]=y[2]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1022 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1023 H2_cont_stat[y[0].split("_")[1]]+=y[2]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1024 break

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1025 #detect contaminations

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1026 #print(H1_cont_stat)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1027 #print(H2_cont_stat)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1028 H1_cont_stat_list=[x for x in H1_cont_stat if H1_cont_stat[x]>0.2]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1029 H2_cont_stat_list=[x for x in H2_cont_stat if H2_cont_stat[x]>0.2]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1030 contamination_H=""

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1031 if len(H1_cont_stat_list)>1 or len(H2_cont_stat_list)>1:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1032 contamination_H="potential contamination from H antigen signals"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1033 elif len(H2_cont_stat_list)==1 and fljB_contig=="NA":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1034 contamination_H="potential contamination from H antigen signals, uncommon weak fljB signals detected"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1035 #get additional antigens

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1036 """

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1037 if ("O-9,46_wbaV" in O_list or "O-9,46_wbaV-from-II-9,12:z29:1,5-SRR1346254" in O_list) and O_list_less_contamination[0].startswith("O-9,"):#not sure should use and float(O9_wbaV)/float(num_1) > 0.1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1038 if "O-9,46_wzy" in O_list:#and float(O946_wzy)/float(num_1) > 0.1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1039 O_choice="O-9,46"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1040 #print "$$$Most possilble Otype: O-9,46"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1041 elif "O-9,46,27_partial_wzy" in O_list:#and float(O94627)/float(num_1) > 0.1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1042 O_choice="O-9,46,27"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1043 #print "$$$Most possilble Otype: O-9,46,27"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1044 elif ("O-3,10_wzx" in O_list) and ("O-9,46_wzy" in O_list) and (O_list[0].startswith("O-3,10") or O_list_less_contamination[0].startswith("O-9,46_wzy")):#and float(O310_wzx)/float(num_1) > 0.1 and float(O946_wzy)/float(num_1) > 0.1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1045 if "O-3,10_not_in_1,3,19" in O_list:#and float(O310_no_1319)/float(num_1) > 0.1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1046 O_choice="O-3,10"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1047 #print "$$$Most possilble Otype: O-3,10 (contain O-3,10_not_in_1,3,19)"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1048 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1049 O_choice="O-1,3,19"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1050 #print "$$$Most possilble Otype: O-1,3,19 (not contain O-3,10_not_in_1,3,19)"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1051 ### end of special test for O9,46 and O3,10 family

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1052

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1053 if O_choice=="O-9,46,27" or O_choice=="O-3,10" or O_choice=="O-1,3,19":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1054 if len(Otypes_uniq)>2:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1055 contamination_O="potential contamination from O antigen signals"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1056 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1057 if len(Otypes_uniq)>1:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1058 if O_choice=="O-4" and len(Otypes_uniq)==2 and "O-9,46,27" in Otypes_uniq: #for special 4,12,27 case such as Bredeney and Schwarzengrund

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1059 contamination_O=""

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1060 elif O_choice=="O-9,46" and len(Otypes_uniq)==2 and "O-9,46_wbaV" in Otypes_uniq and "O-9,46_wzy" in Otypes_uniq: #for special 4,12,27 case such as Bredeney and Schwarzengrund

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1061 contamination_O=""

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1062 """

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1063 additonal_antigents=[]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1064 #print(contamination_O)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1065 #print(contamination_H)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1066 log_file.write(contamination_O+"\n")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1067 log_file.write(contamination_H+"\n")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1068 log_file.close()

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1069 return O_choice,fliC_choice,fljB_choice,special_gene_list,contamination_O,contamination_H,Otypes_uniq,H1_cont_stat_list,H2_cont_stat_list

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1070

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1071 def get_input_K(input_file,lib_dict,data_type,k_size):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1072 #kmer mode; get input_Ks from dict and data_type

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1073 kmers = []

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1074 for h in lib_dict:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1075 kmers += lib_dict[h]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1076 if data_type == '4':

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1077 input_Ks = target_multifasta_kmerizer(input_file, k_size, set(kmers))

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1078 elif data_type == '1' or data_type == '2' or data_type == '3':#set it for now, will change later

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1079 input_Ks = target_read_kmerizer(input_file, k_size, set(kmers))

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1080 elif data_type == '5':#minion_2d_fasta

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1081 #input_Ks = minion_fasta_kmerizer(input_file, k_size, set(kmers))

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1082 input_Ks = target_multifasta_kmerizer(input_file, k_size, set(kmers)) #ed_SL_08172020: change for nanopore workflow

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1083 if data_type == '6':#minion_2d_fastq

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1084 input_Ks = minion_fastq_kmerizer(input_file, k_size, set(kmers))

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1085 return input_Ks

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1086

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1087 def get_kmer_dict(lib_dict,input_Ks):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1088 #kmer mode; get predicted types

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1089 O_dict = {}

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1090 H_dict = {}

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1091 Special_dict = {}

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1092 for h in lib_dict:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1093 score = (len(lib_dict[h] & input_Ks) / len(lib_dict[h])) * 100

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1094 if score > 1: # Arbitrary cut-off for similarity score very low but seems necessary to detect O-3,10 in some cases

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1095 if h.startswith('O-') and score > 25:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1096 O_dict[h] = score

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1097 if h.startswith('fl') and score > 40:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1098 H_dict[h] = score

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1099 if (h[:2] != 'fl') and (h[:2] != 'O-'):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1100 Special_dict[h] = score

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1101 return O_dict,H_dict,Special_dict

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1102

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1103 def call_O_and_H_type(O_dict,H_dict,Special_dict,make_dir):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1104 log_file=open("SeqSero_log.txt","a")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1105 log_file.write("O_scores:\n")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1106 #call O:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1107 highest_O = '-'

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1108 if len(O_dict) == 0:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1109 pass

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1110 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1111 for x in O_dict:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1112 log_file.write(x+"\t"+str(O_dict[x])+"\n")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1113 if ('O-9,46_wbaV__1002' in O_dict and O_dict['O-9,46_wbaV__1002']>70) or ("O-9,46_wbaV-from-II-9,12:z29:1,5-SRR1346254__1002" in O_dict and O_dict['O-9,46_wbaV-from-II-9,12:z29:1,5-SRR1346254__1002']>70): # not sure should use and float(O9_wbaV)/float(num_1) > 0.1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1114 #if 'O-9,46_wzy__1191' in O_dict or "O-9,46_wzy_partial__216" in O_dict: # and float(O946_wzy)/float(num_1) > 0.1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1115 #modified to fix miscall of O-9,46

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1116 if ('O-9,46_wzy__1191' in O_dict and O_dict['O-9,46_wzy__1191']>40) or ("O-9,46_wzy_partial__216" in O_dict and O_dict["O-9,46_wzy_partial__216"]>40): # and float(O946_wzy)/float(num_1) > 0.1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1117 highest_O = "O-9,46"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1118 elif "O-9,46,27_partial_wzy__1019" in O_dict: # and float(O94627)/float(num_1) > 0.1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1119 highest_O = "O-9,46,27"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1120 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1121 highest_O = "O-9" # next, detect O9 vs O2?

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1122 O2 = 0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1123 O9 = 0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1124 for z in Special_dict:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1125 if "tyr-O-9" in z:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1126 O9 = float(Special_dict[z])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1127 if "tyr-O-2" in z:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1128 O2 = float(Special_dict[z])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1129 if O2 > O9:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1130 highest_O = "O-2"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1131 elif ("O-3,10_wzx__1539" in O_dict) and (

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1132 "O-9,46_wzy__1191" in O_dict

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1133 ): # and float(O310_wzx)/float(num_1) > 0.1 and float(O946_wzy)/float(num_1) > 0.1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1134 if "O-3,10_not_in_1,3,19__1519" in O_dict: # and float(O310_no_1319)/float(num_1) > 0.1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1135 highest_O = "O-3,10"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1136 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1137 highest_O = "O-1,3,19"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1138 ### end of special test for O9,46 and O3,10 family

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1139 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1140 try:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1141 max_score = 0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1142 for x in O_dict:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1143 if float(O_dict[x]) >= max_score:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1144 max_score = float(O_dict[x])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1145 #highest_O = x.split("_")[0]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1146 # ed_SL_12182019: modified to fix the O-9,46 error example1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1147 if (x == 'O-9,46_wbaV__1002' or x == 'O-9,46_wbaV-from-II-9,12:z29:1,5-SRR1346254__1002') and ('O-9,46_wzy__1191' not in O_dict and 'O-9,46_wzy_partial__216' not in O_dict):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1148 highest_O = "O-9"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1149 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1150 highest_O = x.split("_")[0]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1151 if highest_O == "O-1,3,19":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1152 highest_O = '-'

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1153 max_score = 0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1154 for x in O_dict:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1155 if x == 'O-1,3,19_not_in_3,10__130':

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1156 pass

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1157 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1158 if float(O_dict[x]) >= max_score:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1159 max_score = float(O_dict[x])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1160 #highest_O = x.split("_")[0]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1161 # ed_SL_12182019: modified to fix the O-9,46 error example1

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1162 if (x == 'O-9,46_wbaV__1002' or x == 'O-9,46_wbaV-from-II-9,12:z29:1,5-SRR1346254__1002') and ('O-9,46_wzy__1191' not in O_dict and 'O-9,46_wzy_partial__216' not in O_dict):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1163 highest_O = "O-9"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1164 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1165 highest_O = x.split("_")[0]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1166 except:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1167 pass

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1168 #call_fliC:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1169 if len(H_dict)!=0:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1170 highest_H_score_both_BC=H_dict[max(H_dict.keys(), key=(lambda k: H_dict[k]))] #used to detect whether fljB existed or not

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1171 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1172 highest_H_score_both_BC=0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1173 highest_fliC = '-'

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1174 highest_fliC_raw = '-'

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1175 highest_Score = 0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1176 log_file.write("\nH_scores:\n")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1177 for s in H_dict:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1178 log_file.write(s+"\t"+str(H_dict[s])+"\n")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1179 if s.startswith('fliC'):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1180 if float(H_dict[s]) > highest_Score:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1181 highest_fliC = s.split('_')[1]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1182 highest_fliC_raw = s

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1183 highest_Score = float(H_dict[s])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1184 #call_fljB

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1185 highest_fljB = '-'

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1186 highest_fljB_raw = '-'

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1187 highest_Score = 0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1188 for s in H_dict:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1189 if s.startswith('fljB'):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1190 if float(H_dict[s]) > highest_Score and float(H_dict[s]) > highest_H_score_both_BC * 0.65: #fljB is special, so use highest_H_score_both_BC to give a general estimate of coverage, currently 0.65 seems pretty good; the reason use a high (0.65) is some fliC and fljB shared with each other

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1191 #highest_fljB = s.split('_')[1]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1192 #highest_fljB_raw = s

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1193 #highest_Score = float(H_dict[s])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1194 if s.split('_')[1]!=highest_fliC:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1195 highest_fljB = s.split('_')[1]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1196 highest_fljB_raw = s

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1197 highest_Score = float(H_dict[s])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1198 log_file.write("\nSpecial_scores:\n")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1199 for s in Special_dict:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1200 log_file.write(s+"\t"+str(Special_dict[s])+"\n")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1201 log_file.close()

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1202 return highest_O,highest_fliC,highest_fljB

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1203

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1204 def get_temp_file_names(for_fq,rev_fq):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1205 #seqsero2 -a; get temp file names

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1206 sam=for_fq+".sam"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1207 bam=for_fq+".bam"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1208 sorted_bam=for_fq+"_sorted.bam"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1209 mapped_fq1=for_fq+"_mapped.fq"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1210 mapped_fq2=rev_fq+"_mapped.fq"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1211 combined_fq=for_fq+"_combined.fq"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1212 for_sai=for_fq+".sai"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1213 rev_sai=rev_fq+".sai"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1214 return sam,bam,sorted_bam,mapped_fq1,mapped_fq2,combined_fq,for_sai,rev_sai

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1215

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1216 def map_and_sort(threads,database,fnameA,fnameB,sam,bam,for_sai,rev_sai,sorted_bam,mapping_mode):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1217 #seqsero2 -a; do mapping and sort

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1218 print("building database...")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1219 subprocess.check_call("bwa index "+database+ " 2>> data_log.txt",shell=True)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1220 print("mapping...")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1221 if mapping_mode=="mem":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1222 subprocess.check_call("bwa mem -k 17 -t "+threads+" "+database+" "+fnameA+" "+fnameB+" > "+sam+ " 2>> data_log.txt",shell=True)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1223 elif mapping_mode=="sam":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1224 if fnameB!="":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1225 subprocess.check_call("bwa aln -t "+threads+" "+database+" "+fnameA+" > "+for_sai+ " 2>> data_log.txt",shell=True)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1226 subprocess.check_call("bwa aln -t "+threads+" "+database+" "+fnameB+" > "+rev_sai+ " 2>> data_log.txt",shell=True)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1227 subprocess.check_call("bwa sampe "+database+" "+for_sai+" "+ rev_sai+" "+fnameA+" "+fnameB+" > "+sam+ " 2>> data_log.txt",shell=True)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1228 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1229 subprocess.check_call("bwa aln -t "+threads+" "+database+" "+fnameA+" > "+for_sai+ " 2>> data_log.txt",shell=True)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1230 subprocess.check_call("bwa samse "+database+" "+for_sai+" "+for_fq+" > "+sam)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1231 subprocess.check_call("samtools view -@ "+threads+" -F 4 -Sh "+sam+" > "+bam,shell=True)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1232 ### check the version of samtools then use differnt commands

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1233 samtools_version=subprocess.Popen(["samtools"],stdout=subprocess.PIPE,stderr=subprocess.PIPE)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1234 out, err = samtools_version.communicate()

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1235 version = str(err).split("ersion:")[1].strip().split(" ")[0].strip()

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1236 print("check samtools version:",version)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1237 ### end of samtools version check and its analysis

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1238 if LooseVersion(version)<=LooseVersion("1.2"):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1239 subprocess.check_call("samtools sort -@ "+threads+" -n "+bam+" "+fnameA+"_sorted",shell=True)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1240 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1241 subprocess.check_call("samtools sort -@ "+threads+" -n "+bam+" >"+sorted_bam,shell=True)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1242

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1243 def extract_mapped_reads_and_do_assembly_and_blast(current_time,sorted_bam,combined_fq,mapped_fq1,mapped_fq2,threads,fnameA,fnameB,database,mapping_mode,phred_offset):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1244 #seqsero2 -a; extract, assembly and blast

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1245 #subprocess.check_call("bamToFastq -i "+sorted_bam+" -fq "+combined_fq,shell=True)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1246 subprocess.check_call("samtools bam2fq "+sorted_bam+" > "+combined_fq+" 2>> data_log.txt",shell=True) ## change to samtools bam2fq. 202509

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1247 #print("fnameA:",fnameA)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1248 #print("fnameB:",fnameB)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1249 if fnameB!="":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1250 #subprocess.check_call("bamToFastq -i "+sorted_bam+" -fq "+mapped_fq1+" -fq2 "+mapped_fq2 + " 2>> data_log.txt",shell=True)#2> /dev/null if want no output

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1251 subprocess.check_call("samtools bam2fq -1 "+mapped_fq1+" -2 "+mapped_fq2+" -0 /dev/null -s /dev/null -n "+sorted_bam+" 2>> data_log.txt",shell=True) ## change to samtools bam2fq. 202509

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1252 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1253 pass

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1254 outdir=current_time+"_temp"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1255 print("assembling...")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1256 if int(threads)>4:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1257 t="4"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1258 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1259 t=threads

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1260 if os.path.getsize(combined_fq)>100 and (fnameB=="" or os.path.getsize(mapped_fq1)>100):#if not, then it's "-:-:-"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1261 if phred_offset == 'auto':

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1262 phred_offset = ''

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1263 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1264 phred_offset = '--phred-offset ' + phred_offset

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1265

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1266 if fnameB!="":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1267 #print("spades.py --careful "+phred_offset+" --pe1-s "+combined_fq+" --pe1-1 "+mapped_fq1+" --pe1-2 "+mapped_fq2+" -t "+t+" -o "+outdir+ " >> data_log.txt 2>&1")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1268 subprocess.check_call("spades.py --careful "+phred_offset+" --pe1-s "+combined_fq+" --pe1-1 "+mapped_fq1+" --pe1-2 "+mapped_fq2+" -t "+t+" -o "+outdir+ " >> data_log.txt 2>&1",shell=True)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1269 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1270 subprocess.check_call("spades.py --careful "+phred_offset+" --pe1-s "+combined_fq+" -t "+t+" -o "+outdir+ " >> data_log.txt 2>&1",shell=True)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1271 new_fasta=fnameA+"_"+database+"_"+mapping_mode+".fasta"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1272 #new_fasta=fnameA+"_"+database.split('/')[-1]+"_"+mapping_mode+".fasta" # change path to databse for packaging

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1273 subprocess.check_call("mv "+outdir+"/contigs.fasta "+new_fasta+ " 2> /dev/null",shell=True)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1274 #os.system("mv "+outdir+"/scaffolds.fasta "+new_fasta+ " 2> /dev/null") contigs.fasta

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1275 subprocess.check_call("rm -rf "+outdir+ " 2> /dev/null",shell=True)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1276 print("blasting...","\n")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1277 xmlfile="blasted_output.xml"#fnameA+"-extracted_vs_"+database+"_"+mapping_mode+".xml"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1278 subprocess.check_call('makeblastdb -in '+new_fasta+' -out '+new_fasta+'_db '+'-dbtype nucl >> data_log.txt 2>&1',shell=True) #temp.txt is to forbid the blast result interrupt the output of our program###1/27/2015

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1279 subprocess.check_call("blastn -query "+database+" -db "+new_fasta+"_db -out "+xmlfile+" -outfmt 5 >> data_log.txt 2>&1",shell=True)###1/27/2015; 08272018, remove "-word_size 10"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1280 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1281 xmlfile="NA"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1282 return xmlfile,new_fasta

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1283

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1284 def judge_subspecies(fnameA):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1285 #seqsero2 -a; judge subspecies on just forward raw reads fastq

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1286 salmID_output=subprocess.Popen("SalmID.py -i "+fnameA,shell=True,stdout=subprocess.PIPE,stderr=subprocess.PIPE)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1287 out, err = salmID_output.communicate()

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1288 out=out.decode("utf-8")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1289 file=open("data_log.txt","a")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1290 file.write(out)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1291 file.close()

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1292 salm_species_scores=out.split("\n")[1].split("\t")[6:]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1293 salm_species_results=out.split("\n")[0].split("\t")[6:]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1294 max_score=0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1295 max_score_index=1 #default is 1, means "I"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1296 for i in range(len(salm_species_scores)):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1297 if max_score<float(salm_species_scores[i]):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1298 max_score=float(salm_species_scores[i])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1299 max_score_index=i

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1300 prediction=salm_species_results[max_score_index].split(".")[1].strip().split(" ")[0]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1301 #if float(out.split("\n")[1].split("\t")[4]) > float(out.split("\n")[1].split("\t")[5]): #bongori and enterica compare

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1302 if float(out.split("\n")[1].split("\t")[4]) > 10 and float(out.split("\n")[1].split("\t")[4]) > float(out.split("\n")[1].split("\t")[5]): ## ed_SL_0318: change SalmID_ssp_threshold

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1303 prediction="bongori" #if not, the prediction would always be enterica, since they are located in the later part

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1304 #if max_score<10: ## ed_SL_0318: change SalmID_ssp_threshold

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1305 if max_score<60:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1306 prediction="-"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1307 ## ed_SL_0818: add for enterica

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1308 if float(out.split("\n")[1].split("\t")[5]) > 10 and float(out.split("\n")[1].split("\t")[5]) > float(out.split("\n")[1].split("\t")[4]):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1309 prediction="enterica"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1310 ##

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1311 return prediction

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1312

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1313 def judge_subspecies_Kmer(Special_dict):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1314 #seqsero2 -k;

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1315 max_score=0

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1316 prediction="-" #default should be I

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1317 for x in Special_dict:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1318 #if "mer" in x: ## ed_SL_0318: change ssp_threshold

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1319 if "mer" in x and float(Special_dict[x]) > 60:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1320 if max_score<float(Special_dict[x]):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1321 max_score=float(Special_dict[x])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1322 prediction=x.split("_")[-1].strip()

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1323 if x.split("_")[-1].strip()=="bongori" and float(Special_dict[x])>95:#if bongori already, then no need to test enterica

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1324 prediction="bongori"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1325 break

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1326 return prediction

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1327

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1328 ## ed_SL_11232019: add notes for missing antigen

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1329 def check_antigens(ssp,O_antigen,H1_antigen,H2_antigen,NA_note):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1330 antigen_note = ''

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1331 if ssp != '-':

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1332 if O_antigen != '-' and H1_antigen == '-' and H2_antigen == '-': # O:-:-

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1333 antigen_note = 'H antigens were not detected. This is an atypical result that should be further investigated. Most Salmonella strains have at least fliC, encoding the Phase 1 H antigen, even if it is not expressed. '

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1334 NA_note = ''

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1335 elif O_antigen != '-' and H1_antigen == '-' and H2_antigen != '-': # O:-:H2

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1336 antigen_note = 'fliC was not detected. This is an atypical result that should be further investigated. Most Salmonella strains have fliC, encoding the Phase 1 H antigen, even if it is not expressed. '

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1337 NA_note = ''

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1338 elif O_antigen == '-' and H1_antigen != '-': # -:H1:X

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1339 antigen_note = 'O antigen was not detected. This result may be due to a rough strain that has deleted the rfb region. For raw reads input, the k-mer workflow is sometimes more sensitive than the microassembly workflow in detecting O antigen. Caution should be used with this approach because the k-mer result may be due to low levels of contamination. '

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1340 NA_note = ''

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1341 elif O_antigen == '-' and H1_antigen == '-' and H2_antigen == '-': # -:-:-

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1342 antigen_note = 'No serotype antigens were detected. This is an atypical result that should be further investigated. '

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1343 NA_note = ''

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1344 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1345 antigen_note = 'The input genome cannot be identified as Salmonella. Check the input for taxonomic ID, contamination, or sequencing quality. '

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1346 NA_note = ''

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1347 if ssp == 'enterica':

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1348 antigen_note += 'Subspecies identification of the input genome cannot be definitively determined. '

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1349 NA_note = ''

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1350 # if [O_antigen, H1_antigen, H2_antigen].count('-') >= 2:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1351 # antigen_note = 'No subspecies marker was detected and less than 2 serotype antigens were detected; further, this genome was not identified as Salmonella. This is an atypical result that should be further investigated. '

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1352 # else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1353 # antigen_note = 'No subspecies marker was detected. This genome may not be Salmonella. This is an atypical result that should be further investigated. '

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1354 return (antigen_note,NA_note)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1355

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1356 ## ed_SL_06062020: rename subspecies ID

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1357 subspecies_ID_dir = {'I': 'Salmonella enterica subspecies enterica (subspecies I)',

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1358 'II': 'Salmonella enterica subspecies salamae (subspecies II)',

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1359 'IIIa': 'Salmonella enterica subspecies arizonae (subspecies IIIa)',

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1360 'IIIb': 'Salmonella enterica subspecies diarizonae (subspecies IIIb)',

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1361 'IV': 'Salmonella enterica subspecies houtenae (subspecies IV)',

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1362 'VI': 'Salmonella enterica subspecies indica (subspecies VI)',

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1363 'VII': 'Salmonella enterica subspecies VII (subspecies VII)',

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1364 'bongori': 'Salmonella bongori',

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1365 'enterica': 'Salmonella enterica',

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1366 '-': '-'}

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1367 ##

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1368

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1369 ## ed_SL_08172020: format check for fasta or fastq in nanopore workflow, convert raw reads fastq to fasta

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1370 def format_check(input_file):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1371 line=open(input_file,'r').readline()

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1372 if line.startswith('>'):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1373 output_file = input_file

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1374 elif line.startswith('@'):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1375 input_file_fa = input_file + '.fasta'

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1376 subprocess.check_call("seqtk seq -A "+input_file+" > "+input_file_fa,shell=True)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1377 output_file = input_file_fa

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1378 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1379 print ('please check the format of input files')

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1380 return (output_file)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1381 ##

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1382

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1383 def main():

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1384 #combine SeqSeroK and SeqSero2, also with SalmID

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1385 args = parse_args()

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1386 input_file = args.i

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1387 data_type = args.t

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1388 analysis_mode = args.m

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1389 mapping_mode=args.b

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1390 threads=args.p

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1391 make_dir=args.d

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1392 clean_mode=args.c

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1393 sample_name=args.n

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1394 ingore_header=args.s

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1395 phred_offset=args.phred_offset

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1396 k_size=27 #will change for bug fixing

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1397 dirpath = os.path.abspath(os.path.dirname(os.path.realpath(__file__)))

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1398 ex_dir = os.path.abspath(os.path.join(os.path.dirname(os.path.dirname(__file__)),'seqsero2s_db')) # ed_SL_09152019: add ex_dir for packaging

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1399 seqsero2_db=ex_dir+"/H_and_O_and_specific_genes.fasta" # ed_SL_11092019: change path to database for packaging

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1400 database="H_and_O_and_specific_genes.fasta"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1401 mlst_count=pickle.load(open(ex_dir+"/mlst.pickle", "rb"))

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1402 note="Note: "

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1403 NA_note="This predicted serotype is not in the Kauffman-White scheme. " # ed_SL_09272019: add for new output format

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1404 if len(sys.argv)==1:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1405 subprocess.check_call(dirpath+"/SeqSero2S.py -h",shell=True)#change name of python file

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1406 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1407 request_id = time.strftime("%m_%d_%Y_%H_%M_%S", time.localtime())

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1408 request_id += str(random.randint(1, 10000000))

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1409 if make_dir is None:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1410 make_dir="SeqSero_result_"+request_id

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1411 make_dir=os.path.abspath(make_dir)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1412 if os.path.isdir(make_dir):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1413 pass

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1414 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1415 subprocess.check_call("mkdir -p "+make_dir,shell=True)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1416 subprocess.check_call("ln -f -s "+seqsero2_db+" "+" ".join(input_file)+" "+make_dir,shell=True) # ed_SL_11092019: change path to database for packaging

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1417 #subprocess.check_call("ln -f -s "+dirpath+"/"+database+" "+" ".join(input_file)+" "+make_dir,shell=True) ### use -f option to force the replacement of links, remove -r and use absolute path instead to avoid link issue (use 'type=os.path.abspath' in -i argument).

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1418 ############################begin the real analysis

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1419 if analysis_mode=="a":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1420 if data_type in ["1","2","3"]:#use allele mode

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1421 for_fq,rev_fq=get_input_files(make_dir,input_file,data_type,dirpath)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1422 os.chdir(make_dir)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1423 ###add a function to tell input files

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1424 fnameA=for_fq.split("/")[-1]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1425 fnameB=rev_fq.split("/")[-1]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1426 current_time=time.strftime("%Y_%m_%d_%H_%M_%S", time.localtime())

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1427 sam,bam,sorted_bam,mapped_fq1,mapped_fq2,combined_fq,for_sai,rev_sai=get_temp_file_names(fnameA,fnameB) #get temp files id

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1428 map_and_sort(threads,database,fnameA,fnameB,sam,bam,for_sai,rev_sai,sorted_bam,mapping_mode) #do mapping and sort

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1429

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1430 ### avoid error out when micro assembly fails. ed_SL_03172020

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1431 try:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1432 xmlfile,new_fasta=extract_mapped_reads_and_do_assembly_and_blast(current_time,sorted_bam,combined_fq,mapped_fq1,mapped_fq2,threads,fnameA,fnameB,database,mapping_mode,phred_offset) #extract the mapped reads and do micro assembly and blast

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1433 except (UnboundLocalError, subprocess.CalledProcessError):

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1434 xmlfile="NA"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1435 H1_cont_stat_list=[]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1436 H2_cont_stat_list=[]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1437 ###

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1438 if xmlfile=="NA":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1439 O_choice,fliC_choice,fljB_choice,special_gene_list,contamination_O,contamination_H=("-","-","-",[],"","")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1440 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1441 Final_list=xml_parse_score_comparision_seqsero(xmlfile) #analyze xml and get parsed results

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1442 file=open("data_log.txt","a")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1443 for x in Final_list:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1444 file.write("\t".join(str(y) for y in x)+"\n")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1445 file.close()

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1446 Final_list_passed=[x for x in Final_list if float(x[0].split("_cov_")[1].split("_")[0])>=0.9 and (x[1]>=int(x[0].split("__")[1]) or x[1]>=int(x[0].split("___")[1].split("_")[3]) or x[1]>1000)]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1447 O_choice,fliC_choice,fljB_choice,special_gene_list,contamination_O,contamination_H,Otypes_uniq,H1_cont_stat_list,H2_cont_stat_list=predict_O_and_H_types(Final_list,Final_list_passed,new_fasta) #predict O, fliC and fljB

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1448 subspecies=judge_subspecies(fnameA) #predict subspecies

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1449 ### ed_SL_06062020: correction VIII -> II

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1450 if subspecies == 'VIII':

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1451 subspecies = 'II'

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1452 ### ed_SL_08132020: correction VII -> IV, according to CDC's suggestion

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1453 if subspecies == 'VII':

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1454 subspecies = 'IV'

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1455 note+='SalmID reports this as ssp VII, which has not been formally recognized. '

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1456 ###

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1457 ### ed_SL_08182020: change serotype ouput for genome without definitive subspecies ID

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1458 ssp_pointer = subspecies

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1459 if subspecies == 'enterica':

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1460 subspecies = '-'

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1461 ###

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1462 ###MLST

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1463 #print("MLST using https://github.com/jordanlab/stringMLST")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1464 #print("7-gene MLST...")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1465 mlst_result = stringmlst(for_fq,rev_fq,data_type,ex_dir)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1466 st = mlst_result[0]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1467 alleles = mlst_result[1]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1468 sorted_alleles = sorted(alleles.items())

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1469 try:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1470 st_count = str(mlst_count[st])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1471 except:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1472 st_count = '0'

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1473 subprocess.call("rm H_and_O_and_specific_genes.fasta* *.sra *.bam *.sam *.fastq *.gz *.fq temp.txt "+fnameA+"*_db* 2> /dev/null",shell=True)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1474 ###

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1475 ###output

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1476 predict_form_ss2,predict_sero_ss2,star_ss2,star_line_ss2,claim_ss2=seqsero_from_formula_to_serotypes(O_choice,fliC_choice,fljB_choice,special_gene_list,subspecies,'ss2')

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1477 predict_form,predict_sero,star,star_line,claim=seqsero_from_formula_to_serotypes(O_choice,fliC_choice,fljB_choice,special_gene_list,subspecies,'ss2s')

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1478 claim="" #04132019, disable claim for new report requirement

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1479 contamination_report=""

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1480 H_list=["fliC_"+x for x in H1_cont_stat_list if len(x)>0]+["fljB_"+x for x in H2_cont_stat_list if len(x)>0]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1481 if contamination_O!="" and contamination_H=="":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1482 contamination_report="#Potential inter-serotype contamination detected from O antigen signals. All O-antigens detected:"+"\t".join(Otypes_uniq)+"."

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1483 elif contamination_O=="" and contamination_H!="":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1484 contamination_report="#Potential inter-serotype contamination detected or potential thrid H phase from H antigen signals. All H-antigens detected:"+"\t".join(H_list)+"."

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1485 elif contamination_O!="" and contamination_H!="":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1486 contamination_report="#Potential inter-serotype contamination detected from both O and H antigen signals.All O-antigens detected:"+"\t".join(Otypes_uniq)+". All H-antigens detected:"+"\t".join(H_list)+"."

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1487 if contamination_report!="":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1488 #contamination_report="potential inter-serotype contamination detected (please refer below antigen signal report for details)." #above contamination_reports are for back-up and bug fixing #web-based mode need to be re-used, 04132019

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1489 contamination_report="Co-existence of multiple serotypes detected, indicating potential inter-serotype contamination. See 'Extracted_antigen_alleles.fasta' for detected serotype determinant alleles. "

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1490

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1491 ### ed_SL_11232019: add notes for missing antigen

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1492 if O_choice=="":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1493 O_choice="-"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1494 antigen_note,NA_note=check_antigens(ssp_pointer,O_choice,fliC_choice,fljB_choice,NA_note)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1495 if sample_name:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1496 print ("Sample name:\t"+sample_name)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1497 ###

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1498 if clean_mode:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1499 subprocess.check_call("rm -rf "+make_dir,shell=True)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1500 make_dir="none-output-directory due to '-c' flag"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1501 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1502 new_file=open("SeqSero_result.txt","w")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1503 ### ed_SL_01152020: add new output

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1504 conta_note="yes" if "inter-serotype contamination" in contamination_report else "no"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1505 tsv_file=open("SeqSero_result.tsv","w")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1506 if ingore_header:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1507 pass

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1508 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1509 tsv_file.write("Sample name\tOutput directory\tInput files\tO antigen prediction\tH1 antigen prediction(fliC)\tH2 antigen prediction(fljB)\tPredicted identification\tPredicted antigenic profile\tPredicted serotype\tPredicted serotype (SeqSero2 v1.3.2)\tPotential inter-serotype contamination\tNote\tST\n")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1510 if sample_name:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1511 new_file.write("Sample name:\t"+sample_name+"\n")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1512 tsv_file.write(sample_name+'\t')

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1513 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1514 tsv_file.write(input_file[0].split('/')[-1]+'\t')

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1515 ###

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1516 if "N/A" not in predict_sero:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1517 new_file.write("Output directory:\t"+make_dir+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1518 "Input files:\t"+"\t".join(input_file)+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1519 "O antigen prediction:\t"+O_choice+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1520 "H1 antigen prediction(fliC):\t"+fliC_choice+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1521 "H2 antigen prediction(fljB):\t"+fljB_choice+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1522 "Predicted identification:\t"+subspecies_ID_dir[ssp_pointer]+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1523 "Predicted antigenic profile:\t"+predict_form+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1524 "Predicted serotype:\t"+predict_sero+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1525 "Predicted serotype (SeqSero2 v1.3.2):\t"+predict_sero_ss2+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1526 note+contamination_report+star_line+claim+antigen_note+"\n")#+##

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1527 tsv_file.write(make_dir+"\t"+" ".join(input_file)+"\t"+O_choice+"\t"+fliC_choice+"\t"+fljB_choice+"\t"+subspecies_ID_dir[ssp_pointer]+"\t"+predict_form+"\t"+predict_sero+"\t"+predict_sero_ss2+"\t"+conta_note+"\t"+contamination_report+star_line+claim+antigen_note+"\t"+st+"\n")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1528 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1529 new_file.write("Output directory:\t"+make_dir+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1530 "Input files:\t"+"\t".join(input_file)+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1531 "O antigen prediction:\t"+O_choice+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1532 "H1 antigen prediction(fliC):\t"+fliC_choice+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1533 "H2 antigen prediction(fljB):\t"+fljB_choice+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1534 "Predicted identification:\t"+subspecies_ID_dir[ssp_pointer]+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1535 "Predicted antigenic profile:\t"+predict_form+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1536 "Predicted serotype:\t"+subspecies+' '+predict_form+"\n"+ # add serotype output for "N/A" prediction, add subspecies

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1537 "Predicted serotype (SeqSero2 v1.3.2):\t"+subspecies+' '+predict_form_ss2+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1538 note+NA_note+contamination_report+star_line+claim+antigen_note+"\n")#+##

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1539 tsv_file.write(make_dir+"\t"+" ".join(input_file)+"\t"+O_choice+"\t"+fliC_choice+"\t"+fljB_choice+"\t"+subspecies_ID_dir[ssp_pointer]+"\t"+predict_form+"\t"+subspecies+' '+predict_form+"\t"+subspecies+' '+predict_form_ss2+"\t"+conta_note+"\t"+NA_note+contamination_report+star_line+claim+antigen_note+"\t"+st+"\n")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1540 ##MLST

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1541 new_file.write("Sequence type:\t"+st+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1542 "Number of ST"+st+" strains in EnteroBase:\t"+st_count+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1543 "\n".join([k+":\t"+v for k,v in sorted_alleles]))

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1544 ###

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1545 new_file.close()

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1546 tsv_file.close()

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1547 if "N/A" not in predict_sero:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1548 print("Output directory:\t"+make_dir+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1549 "Input files:\t"+"\t".join(input_file)+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1550 "O antigen prediction:\t"+O_choice+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1551 "H1 antigen prediction(fliC):\t"+fliC_choice+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1552 "H2 antigen prediction(fljB):\t"+fljB_choice+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1553 "Predicted identification:\t"+subspecies_ID_dir[ssp_pointer]+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1554 "Predicted antigenic profile:\t"+predict_form+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1555 "Predicted serotype:\t"+predict_sero+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1556 "Predicted serotype (SeqSero2 v1.3.2):\t"+predict_sero_ss2+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1557 note+contamination_report+star_line+claim+antigen_note+"\n")#+##

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1558 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1559 print("Output directory:\t"+make_dir+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1560 "Input files:\t"+"\t".join(input_file)+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1561 "O antigen prediction:\t"+O_choice+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1562 "H1 antigen prediction(fliC):\t"+fliC_choice+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1563 "H2 antigen prediction(fljB):\t"+fljB_choice+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1564 "Predicted identification:\t"+subspecies_ID_dir[ssp_pointer]+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1565 "Predicted antigenic profile:\t"+predict_form+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1566 "Predicted serotype:\t"+subspecies+' '+predict_form+"\n"+ # add serotype output for "N/A" prediction, subspecies

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1567 "Predicted serotype (SeqSero2 v1.3.2):\t"+subspecies+' '+predict_form_ss2+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1568 note+NA_note+contamination_report+star_line+claim+antigen_note+"\n")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1569 ###MLST

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1570 print("Sequence type: "+st)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1571 print("Number of ST"+st+" strains in EnteroBase: "+st_count)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1572 #print("Allele profile...")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1573 for k,v in sorted_alleles:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1574 print(k+': '+v)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1575 print('\n')

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1576 ###

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1577 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1578 print("Allele modes only support raw reads datatype, i.e. '-t 1 or 2 or 3'; please use '-m k'")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1579 elif analysis_mode=="k":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1580 #ex_dir = os.path.dirname(os.path.realpath(__file__))

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1581 ex_dir = os.path.abspath(os.path.join(os.path.dirname(os.path.dirname(__file__)),'seqsero2s_db')) # ed_SL_09152019: change ex_dir for packaging

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1582 for_fq,rev_fq=get_input_files(make_dir,input_file,data_type,dirpath)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1583 input_file = for_fq #-k will just use forward because not all reads were used

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1584 os.chdir(make_dir)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1585 ### ed_SL_08182020: use assembly workflow for nanopore fastq, convert fastq to fasta

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1586 if data_type == "5":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1587 input_file = format_check(for_fq)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1588 ###

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1589 f = open(ex_dir + '/antigens.pickle', 'rb')

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1590 lib_dict = pickle.load(f)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1591 f.close

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1592 input_Ks=get_input_K(input_file,lib_dict,data_type,k_size)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1593 O_dict,H_dict,Special_dict=get_kmer_dict(lib_dict,input_Ks)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1594 highest_O,highest_fliC,highest_fljB=call_O_and_H_type(O_dict,H_dict,Special_dict,make_dir)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1595 subspecies=judge_subspecies_Kmer(Special_dict)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1596 if subspecies=="IIb" or subspecies=="IIa":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1597 subspecies="II"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1598 ### ed_SL_06062020: correction VIII -> II

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1599 if subspecies == 'VIII':

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1600 subspecies = 'II'

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1601 ### ed_SL_08132020: correction VII -> IV, according to CDC's suggestion

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1602 if subspecies == 'VII':

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1603 subspecies = 'IV'

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1604 note+='SalmID reports this as ssp VII, which has not been formally recognized. '

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1605 ###

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1606 ### ed_SL_08182020: change serotype ouput for genome without definitive subspecies ID

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1607 ssp_pointer = subspecies

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1608 if subspecies == 'enterica':

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1609 subspecies = '-'

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1610 ###

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1611 predict_form_ss2,predict_sero_ss2,star_ss2,star_line_ss2,claim_ss2 = seqsero_from_formula_to_serotypes(

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1612 highest_O.split('-')[1], highest_fliC, highest_fljB, Special_dict,subspecies, 'ss2')

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1613 predict_form,predict_sero,star,star_line,claim = seqsero_from_formula_to_serotypes(

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1614 highest_O.split('-')[1], highest_fliC, highest_fljB, Special_dict,subspecies, 'ss2s')

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1615 claim="" #no claim any more based on new output requirement

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1616

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1617 ### ed_SL_11232019: add notes for missing antigen

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1618 if highest_O.split('-')[-1]=="":

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1619 O_choice="-"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1620 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1621 O_choice=highest_O.split('-')[-1]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1622 antigen_note,NA_note=check_antigens(ssp_pointer,O_choice,highest_fliC,highest_fljB,NA_note)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1623 if sample_name:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1624 print ("Sample name:\t"+sample_name)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1625 ###

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1626 ###MLST

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1627 if data_type in ["4","5"]:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1628 #print("MLST using https://github.com/tseemann/mlst")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1629 #print("7-gene MLST...")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1630 mlst_result = mlst(args.i[0])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1631 if data_type in ["1","2","3"]:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1632 #print("MLST using https://github.com/jordanlab/stringMLST")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1633 #print("7-gene MLST...")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1634 mlst_result = stringmlst(for_fq,rev_fq,data_type,ex_dir)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1635 st = mlst_result[0]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1636 alleles = mlst_result[1]

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1637 sorted_alleles = sorted(alleles.items())

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1638 try:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1639 st_count = str(mlst_count[st])

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1640 except:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1641 st_count = '0'

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1642 subprocess.call("rm *.fasta* *.fastq *.gz *.fq temp.txt *.sra 2> /dev/null",shell=True)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1643 ###

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1644 ###output

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1645 if clean_mode:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1646 subprocess.check_call("rm -rf "+make_dir,shell=True)

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1647 make_dir="none-output-directory due to '-c' flag"

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1648 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1649 new_file=open("SeqSero_result.txt","w")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1650 tsv_file=open("SeqSero_result.tsv","w")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1651 if ingore_header:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1652 pass

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1653 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1654 tsv_file.write("Sample name\tOutput directory\tInput files\tO antigen prediction\tH1 antigen prediction(fliC)\tH2 antigen prediction(fljB)\tPredicted identification\tPredicted antigenic profile\tPredicted serotype\tPredicted serotype (SeqSero2 v1.3.2)\tNote\tST\n")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1655 if sample_name:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1656 new_file.write("Sample name:\t"+sample_name+"\n")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1657 tsv_file.write(sample_name+'\t')

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1658 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1659 tsv_file.write(input_file.split('/')[-1]+'\t')

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1660 ###

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1661 if "N/A" not in predict_sero:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1662 new_file.write("Output directory:\t"+make_dir+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1663 "Input files:\t"+input_file+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1664 "O antigen prediction:\t"+O_choice+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1665 "H1 antigen prediction(fliC):\t"+highest_fliC+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1666 "H2 antigen prediction(fljB):\t"+highest_fljB+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1667 "Predicted identification:\t"+subspecies_ID_dir[ssp_pointer]+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1668 "Predicted antigenic profile:\t"+predict_form+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1669 "Predicted serotype:\t"+predict_sero+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1670 "Predicted serotype (SeqSero2 v1.3.2):\t"+predict_sero_ss2+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1671 note+star_line+claim+antigen_note+"\n")#+##

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1672 tsv_file.write(make_dir+"\t"+input_file+"\t"+O_choice+"\t"+highest_fliC+"\t"+highest_fljB+"\t"+subspecies_ID_dir[ssp_pointer]+"\t"+predict_form+"\t"+predict_sero+"\t"+predict_sero_ss2+"\t"+star_line+claim+antigen_note+"\t"+st+"\n")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1673 else:

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1674 new_file.write("Output directory:\t"+make_dir+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1675 "Input files:\t"+input_file+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1676 "O antigen prediction:\t"+O_choice+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1677 "H1 antigen prediction(fliC):\t"+highest_fliC+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1678 "H2 antigen prediction(fljB):\t"+highest_fljB+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1679 "Predicted identification:\t"+subspecies_ID_dir[ssp_pointer]+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1680 "Predicted antigenic profile:\t"+predict_form+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1681 "Predicted serotype:\t"+subspecies+' '+predict_form+"\n"+ # add serotype output for "N/A" prediction, subspecies

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1682 "Predicted serotype (SeqSero2 v1.3.2):\t"+subspecies+' '+predict_form_ss2+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1683 note+NA_note+star_line+claim+antigen_note+"\n")#+##

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1684 tsv_file.write(make_dir+"\t"+input_file+"\t"+O_choice+"\t"+highest_fliC+"\t"+highest_fljB+"\t"+subspecies_ID_dir[ssp_pointer]+"\t"+predict_form+"\t"+subspecies+' '+predict_form+"\t"+subspecies+' '+predict_form_ss2+"\t"+NA_note+star_line+claim+antigen_note+"\t"+st+"\n")

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1685 ###MLST

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne

parents:

diff changeset

1686 new_file.write("Sequence type:\t"+st+"\n"+

cfc91e1d2c9b planemo upload commit 936a627c4fc706080f07ec678f89e8256a7e7895

jpayne