Mercurial > repos > rliterman > csp2

comparison 1 @ 37:f62b9bcf7cf3

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload"
author rliterman
date Fri, 06 Dec 2024 08:08:22 -0500
parents
children
comparison
equal deleted inserted replaced
36:f6cd698b6f62 37:f62b9bcf7cf3
1 <tool id="csp2-screen" name="CSP2 (Screening Mode)" version="0.9.7_Dev21">
2 <description>Screen query assemblies against reference assemblies</description>
3 <requirements>
4 <requirement type="package" version="24.10.1">nextflow</requirement>
5 <requirement type="package" version="1.5.8">micromamba</requirement>
6 </requirements>
7 <version_command>nextflow -version</version_command>
8 <command detect_errors="aggressive"><![CDATA[
9 export CSP2_DIR=\$PWD;
10 mkdir -p \$CSP2_DIR/queries \$CSP2_DIR/references;
11
12 if [ -n "$query_fasta" ] && [ "$query_fasta" != "None" ]; then
13 #for query in $query_fasta:
14 ln -sf ${query} \$CSP2_DIR/queries/${query.element_identifier};
15 #end for
16 export QUERY_FASTA_ARG="--fasta \$CSP2_DIR/queries";
17 else
18 export QUERY_FASTA_ARG="";
19 fi;
20
21 if [ -n "$query_reads" ] && [ "$query_reads" != "None" ]; then
22 #for query in $query_reads:
23 ln -sf ${query} \$CSP2_DIR/queries/${query.element_identifier};
24 #end for
25 export QUERY_READS_ARG="--reads \$CSP2_DIR/queries";
26 else
27 export QUERY_READS_ARG="";
28 fi;
29
30 if [ -n "$ref_fasta" ] && [ "$ref_fasta" != "None" ]; then
31 #for ref in $ref_fasta:
32 ln -sf ${ref} \$CSP2_DIR/references/${ref.element_identifier};
33 #end for
34 export REF_FASTA_ARG="--ref_fasta \$CSP2_DIR/references";
35 else
36 export REF_FASTA_ARG="";
37 fi;
38
39 if [ -n "$ref_reads" ] && [ "$ref_reads" != "None" ]; then
40 #for ref in $ref_reads:
41 ln -sf ${ref} \$CSP2_DIR/references/${ref.element_identifier};
42 #end for
43 export REF_READS_ARG="--ref_reads \$CSP2_DIR/references";
44 else
45 export REF_READS_ARG="";
46 fi;
47
48 if [ -n "$trim_name" ] && [ "$trim_name" != "None" ]; then
49 export TRIM_ARG="--trim_name $trim_name";
50 else
51 export TRIM_ARG="";
52 fi;
53
54 if [ -n "$ref_id" ] && [ "$ref_id" != "None" ]; then
55 export REF_ID_ARG="--ref_id $ref_id";
56 else
57 export REF_ID_ARG="";
58 fi;
59
60 nextflow run ${__tool_directory__}/CSP2/CSP2.nf -profile csp2_galaxy --runmode screen \$QUERY_FASTA_ARG \$REF_FASTA_ARG \$QUERY_READS_ARG \$REF_READS_ARG \$REF_ID_ARG \$TRIM_ARG --readext $readext --forward $forward --reverse $reverse --ref_readext $readext --ref_forward $forward --ref_reverse $reverse --min_cov $min_cov --min_iden $min_iden --min_len $min_len --ref_edge $ref_edge --query_edge $query_edge --dwin $dwin --wsnps $wsnps --out \$CSP2_DIR/CSP2_Screen_Output > Nextflow_Log.txt 2>&1;
61 ]]>
62 </command>
63 <inputs>
64 <param name="query_fasta" type="data" format="fasta" value="" label="Query assemblies" multiple="true" optional="true" />
65 <param name="ref_fasta" type="data" format="fasta" value="" label="Reference assemblies" multiple="true" optional="true" />
66 <param name="query_reads" type="data" format="fastq,fastq.gz" value="" label="Query reads" multiple="true" optional="true" />
67 <param name="ref_reads" type="data" format="fastq,fastq.gz" value="" label="Reference reads" multiple="true" optional="true" />
68 <param name="min_cov" type="float" value="85" label="Minimum reference genome coverage to proceed with distance estimation" optional="true" />
69 <param name="min_iden" type="float" value="99" label="Minimum alignment percent identity to detect SNPs" optional="true" />
70 <param name="min_len" type="integer" value="500" label="Minimum alignment length to detect SNPs" optional="true" />
71 <param name="ref_edge" type="integer" value="150" label="Prune SNPs within this many bases of reference contig edge" optional="true" />
72 <param name="query_edge" type="integer" value="150" label="Prune SNPs within this many bases of query contig edge" optional="true" />
73 <param name="dwin" type="text" value="1000,125,15" label="Comma-separated set of window sizes for SNP density filtration (Set to 0 to disable density filtration)" optional="true" />
74 <param name="wsnps" type="text" value="3,2,1" label="Comma-separated list of maximum SNP counts per density window" optional="true" />
75 <param name="readext" type="text" value="fastq.gz" label="Read extension format (e.g., fastq.gz)" optional="true" />
76 <param name="forward" type="text" value="_1.fastq.gz" label="Forward read suffix (e.g. _1.fastq.gz)" optional="true" />
77 <param name="reverse" type="text" value="_2.fastq.gz" label="Forward read suffix (e.g. _2.fastq.gz)" optional="true" />
78 <param name="trim_name" type="text" value="" label="Text to remove from all file names (e.g., _contigs_skesa)" optional="true" />
79 <param name="ref_id" type="text" value="" label="Comma-separated list of desired Reference IDs (e.g., Sample_A,Sample_B)" optional="true" />
80 </inputs>
81 <outputs>
82 <data name="raw_mummer" format="tabular" label="Raw MUMmer Output" from_work_dir="CSP2_Screen_Output/Raw_MUMmer_Summary.tsv" />
83 <data name="isolate_data" format="tabular" label="Isolate Data" from_work_dir="CSP2_Screen_Output/Isolate_Data.tsv" />
84 <data name="screening_results" format="tabular" label="Screening Results" from_work_dir="CSP2_Screen_Output/Screening_Results.tsv" />
85 <data name="nextflow_log" format="txt" label="Nextflow Log" from_work_dir="Nextflow_Log.txt" />
86 </outputs>
87 <tests>
88 <test>
89 <param name="query_fasta">
90 <collection type="list">
91 <element name="Sample_A" value="assemblies/Sample_A.fasta" />
92 <element name="Sample_B" value="assemblies/Sample_B.fasta" />
93 <element name="Sample_C" value="assemblies/Sample_C.fasta" />
94 <element name="Sample_D" value="assemblies/Sample_D.fasta" />
95 <element name="Sample_E" value="assemblies/Sample_E.fasta" />
96 <element name="Sample_F" value="assemblies/Sample_F.fasta" />
97 <element name="Sample_G" value="assemblies/Sample_G.fasta" />
98 <element name="Sample_H" value="assemblies/Sample_H.fasta" />
99 <element name="Sample_I" value="assemblies/Sample_I.fasta" />
100 <element name="Sample_J" value="assemblies/Sample_J.fasta" />
101 <element name="Sample_K" value="assemblies/Sample_K.fasta" />
102 <element name="Sample_L" value="assemblies/Sample_L.fasta" />
103 <element name="Sample_M" value="assemblies/Sample_M.fasta" />
104 <element name="Sample_N" value="assemblies/Sample_N.fasta" />
105 <element name="Sample_O" value="assemblies/Sample_O.fasta" />
106 </collection>
107 </param>
108 <param name="query_reads">
109 <collection type="list">
110 <element name="Forward" value="reads/Week_42_Reads_1.fq.gz" />
111 <element name="Reverse" value="reads/Week_42_Reads_2.fq.gz" />
112 </collection>
113 </param>
114
115 <param name="ref_id" value="Sample_A,Sample_B" />
116 <param name="readext" value="fq.gz" />
117 <param name="forward" value="_1.fq.gz" />
118 <param name="reverse" value="_2.fq.gz" />
119
120 <output name="screening_results" value="Screening_Results.tsv" />
121 <output name="isolate_data" value="Isolate_Data.tsv" />
122 </test>
123 </tests>
124 <help>
125 This tool takes query assemblies and reference assemblies and calculates the pairwise distance between each query/reference combination. If no reference is provided, all queries are compared to all other queries.
126 </help>
127 <citations>
128 <citation type="doi">10.XXXX/placeholder.doi</citation>
129 <citation type="bibtex">@article{example2024,title={CFSAN SNP Pipeline 2 (CSP2): a pipeline for fast and accurate SNP distance estimation from bacterial genome assemblies.},author={Doe, John and Smith, Jane},journal={Submitted},year={2024}}
130 </citation>
131 </citations>
132 </tool>
133