kkonganti@0: #!/usr/bin/env python3
kkonganti@0: 
kkonganti@0: import os
kkonganti@0: import re
kkonganti@0: import glob
kkonganti@0: import argparse
kkonganti@0: import logging
kkonganti@0: 
kkonganti@0: def main():
kkonganti@0:     # READ IN ARGUMENTS
kkonganti@0:     desc = """
kkonganti@0:     Takes in a file with flowcell ID, one per line and creates soft links
kkonganti@0:     to 'fastq_pass' directory at target location.
kkonganti@0: 
kkonganti@0:     Ex:
kkonganti@0: 
kkonganti@0:     prepare_nanopore_fastq_dir.py \
kkonganti@0:         -o /hpc/scratch/Kranti.Konganti/np_test \
kkonganti@0:         -f flowcells.txt
kkonganti@0: 
kkonganti@0:     where flowcells.txt contains the following lines:
kkonganti@0: 
kkonganti@0:     FAL11127
kkonganti@0:     FAL11151
kkonganti@0: 
kkonganti@0:     """
kkonganti@0:     parser = argparse.ArgumentParser(prog='prepare_nanopore_fastq_dir.py',
kkonganti@0:         formatter_class=argparse.ArgumentDefaultsHelpFormatter,
kkonganti@0:         description=desc)
kkonganti@0:     required = parser.add_argument_group('required arguments')
kkonganti@0:     
kkonganti@0:     required.add_argument("-f", dest='flowcells', required=True,
kkonganti@0:         help="Path to a text file containing Nanopore flowcell IDs, one per line")
kkonganti@0:     required.add_argument("-i", dest='inputdir',
kkonganti@0:         required=False, action='append', nargs='*',
kkonganti@0:         help="Path to search directory. This directory location is where" +
kkonganti@0:             " the presence of 'fastq_pass' will be searched for each flowcell.")
kkonganti@0:     required.add_argument("-o", dest='outputdir',
kkonganti@0:         required=True,
kkonganti@0:         help="Path to output directory. This directory is created by the script" +
kkonganti@0:             " and new soft links (symlinks) are created in this directory.")
kkonganti@0:     
kkonganti@0:     args = parser.parse_args()
kkonganti@0:     flowcells = args.flowcells
kkonganti@0:     output = args.outputdir
kkonganti@0:     inputs = args.inputdir
kkonganti@0: 
kkonganti@0:     logging.basicConfig(format='%(asctime)s - %(levelname)s => %(message)s', level=logging.DEBUG)
kkonganti@0: 
kkonganti@0:     if not inputs:
kkonganti@0:         inputs = ['/projects/nanopore/raw']
kkonganti@0:         nanopore_machines = ['RazorCrest', 'Revolution', 'ObiWan', 'MinIT',
kkonganti@0:             'Mayhem', 'CaptainMarvel', 'MinION', 'MinION_Padmini', 'RogueOne']
kkonganti@0:         logging.info(f"Searching default path(s). Use -i option if custom path should be searched.")
kkonganti@0:     else:
kkonganti@0:         nanopore_machines = ['custom']
kkonganti@0: 
kkonganti@0:     fastq_pass_found = {}
kkonganti@0:     was_fastq_pass_found = []
kkonganti@0: 
kkonganti@0:     for each_input in inputs:
kkonganti@0:         for machine in nanopore_machines:
kkonganti@0:             if ''.join(nanopore_machines) != 'custom':
kkonganti@0:                 input = os.path.join(each_input, machine)
kkonganti@0:             else:
kkonganti@0:                 input = ''.join(each_input)
kkonganti@0: 
kkonganti@0:             logging.info(f"Searching path: {input}")
kkonganti@0: 
kkonganti@0:             if (os.path.exists(flowcells) and os.path.getsize(flowcells) > 0):
kkonganti@0:                 with open(flowcells, 'r') as fcells:
kkonganti@0:                     for flowcell in fcells:
kkonganti@0:                         if re.match('^\s*$', flowcell):
kkonganti@0:                             continue
kkonganti@0:                         flowcell = flowcell.strip()
kkonganti@0:                         fastq_pass_path = glob.glob(os.path.join(input, flowcell, f"**", f"*[!fast5]*", 'fastq_pass'))
kkonganti@0:                         # Try one more time since the flowcell user is trying to query may be the parent directory
kkonganti@0:                         # of fastq_pass
kkonganti@0:                         fastq_pass = fastq_pass_path if fastq_pass_path else glob.glob(os.path.join(input, f"**", f"*[!fast5]*", flowcell, 'fastq_pass'))
kkonganti@0:                         if not fastq_pass:
kkonganti@0:                             # logging.warning(f"Flowcell " +
kkonganti@0:                             #     os.path.join(input, flowcell).strip() +
kkonganti@0:                             #     f" does not seem to have a fastq_pass directory! Skipped!!")
kkonganti@0:                             if not flowcell in fastq_pass_found.keys():
kkonganti@0:                                 fastq_pass_found[flowcell] = 0 
kkonganti@0:                         else:
kkonganti@0:                             fastq_pass_found[flowcell] = 1
kkonganti@0:                             sym_link_dir = os.path.join(output, flowcell)
kkonganti@0:                             sym_link_dir_dest = os.path.join(sym_link_dir, 'fastq_pass')
kkonganti@0:                             if not os.path.exists(sym_link_dir):
kkonganti@0:                                 os.makedirs(sym_link_dir)
kkonganti@0:                                 os.symlink(
kkonganti@0:                                     ''.join(fastq_pass),
kkonganti@0:                                     sym_link_dir_dest, target_is_directory=True
kkonganti@0:                                 )
kkonganti@0:                                 logging.info(f"New soft link created: {sym_link_dir_dest}")
kkonganti@0:                             else:
kkonganti@0:                                 logging.info(f"Soft link {sym_link_dir_dest} already exists! Skipped!!")
kkonganti@0:                     fcells.close()
kkonganti@0:             else:
kkonganti@0:                 logging.error(f"File {flowcells} is empty or does not exist!\n")
kkonganti@0: 
kkonganti@0:     for k,v in fastq_pass_found.items():
kkonganti@0:         if not v:
kkonganti@0:             was_fastq_pass_found.append(k)
kkonganti@0: 
kkonganti@0:     if was_fastq_pass_found:
kkonganti@0:         logging.warning("Did not find fastq_pass folder for the supplied flowcells: " +
kkonganti@0:                 ', '.join(was_fastq_pass_found))
kkonganti@0: 
kkonganti@0:     if was_fastq_pass_found and len(was_fastq_pass_found) == len(fastq_pass_found):
kkonganti@0:         logging.error(f"None of the supplied flowcells were found! The output directory, {output} may not have been created!")
kkonganti@0:     else:
kkonganti@0:         logging.info(f"NOTE: Now you can use {output} directory as --input to cpipes.\n")
kkonganti@0: 
kkonganti@0: if __name__ == "__main__":
kkonganti@0:     main()