view edena_ovl_wrapper.py @ 1:cd6cc6d76708 draft

Simplify passing repeated params to Python script. Add more info to help sections.
author crs4
date Fri, 18 Oct 2013 14:09:11 -0400
parents 60609a9cef3b
children
line wrap: on
line source

# -*- coding: utf-8 -*-
"""
Edena (overlapping)
version 0.2.1 (andrea.pinna@crs4.it)
"""

import optparse
import shutil
import subprocess
import sys

def __main__():
    # load arguments
    print 'Parsing Edena (overlapping) input options...'
    parser = optparse.OptionParser()
    parser.add_option('--unpaired_input', action='append', dest='unpaired_input', help='')
    parser.add_option('--dr_pair_1', action='append', dest='dr_pair_1', help='')
    parser.add_option('--dr_pair_2', action='append', dest='dr_pair_2', help='')
    parser.add_option('--rd_pair_1', action='append', dest='rd_pair_1', help='')
    parser.add_option('--rd_pair_2', action='append', dest='rd_pair_2', help='')
    parser.add_option('--nThreads', dest='nThreads', type='int', help='')
    parser.add_option('--minOlap', dest='minOlap', type='int', help='')
    parser.add_option('--readsTruncation', dest='readsTruncation', type='int', help='')
    parser.add_option('--output', dest='output', help='')
    parser.add_option('--logfile', dest='logfile', help='')
    (options, args) = parser.parse_args()
    if len(args) > 0:
        parser.error('Wrong number of arguments')

    # build Edena (overlapping) command to be executed
    # unpaired input(s)
    if options.unpaired_input:
        unpaired_input = '-r'
        for item in options.unpaired_input:
            unpaired_input += ' %s' % (item)
    else:
        unpaired_input = ''
    # direct-reverse paired-end files
    if options.dr_pair_1 and options.dr_pair_2:
        dr_pairs = '-DRpairs'
        for i in range(len(options.dr_pair_1)):
            dr_pairs += ' %s %s' % (options.dr_pair_1[i], options.dr_pair_2[i])
    else:
        dr_pairs = ''
     # reverse-direct paired-end files
    if options.rd_pair_1 and options.rd_pair_2:
        rd_pairs = '-RDpairs'
        for i in range(len(options.rd_pair_1)):
            rd_pairs += ' %s %s' % (options.rd_pair_1[i], options.rd_pair_2[i])
    else:
        rd_pairs = ''
    # nThreads
    nThreads = '-nThreads %d' % (options.nThreads) if options.nThreads is not None else ''
    # minimum overlap
    minOlap = '-M %d' % (options.minOlap) if options.minOlap is not None else ''
    # 3' end reads truncation
    readsTruncation = '-t %d' % (options.readsTruncation) if options.readsTruncation is not None else ''
    # output file(s)
    output = options.output
    logfile = options.logfile

    # Build Edena (overlapping) command
    cmd = 'edena %s %s %s %s %s %s -p galaxy_output' % (unpaired_input, dr_pairs, rd_pairs, nThreads, minOlap, readsTruncation)
    print '\nEdena (overlapping) command to be executed:\n %s' % (cmd)

    # Execution of Edena
    print 'Executing Edena (overlapping)...'
    log = open(logfile, 'w') if logfile else sys.stdout
    try:
        subprocess.check_call(cmd, stdout=log, stderr=subprocess.STDOUT, shell=True) # need to redirect stderr because edena writes some logging info there (e.g. "Computing overlaps >=30...")
    finally:
        if log != sys.stdout:
            log.close()
    print 'Edena (overlapping) executed!'

    shutil.move('galaxy_output.ovl', output)


if __name__ == "__main__":
    __main__()