view SMART/DiffExpAnal/wrappGSNAP.py @ 18:94ab73e8a190

Uploaded

#! /usr/bin/env python

import os, sys, subprocess,tempfile
from optparse import OptionParser

def stop_err(msg):
    sys.stderr.write('%s\n' % msg)
    sys.exit()

def __main__():
    #Parse Command Line
    description = "GMAP/GSNAP version:2012-12-20."
    parser = OptionParser(description = description)
    parser.add_option("-d", "--genomeName", dest="genomeName", help="Define the reference genome name.[compulsory]")
    parser.add_option("-o", "--outputFile", dest="outputfile", help="output[compulsory]")
    #parser.add_option("-D", "--workingDir", dest="workingdir", help="Define the directory of writing reference genome index.[compulsory]")
    parser.add_option("-k", "--kmer", dest="kmer", default=12, help="Choose kmer value (<=16), a big kmer value can take more RAM(4Go).[compulsory]")
    parser.add_option("-i", "--inputFasta", dest="inputFastaFile", help="Reference genome file, fasta format.[compulsory]")
    parser.add_option("-q", "--inputFastq", dest="inputFastqFile", help="Input fastq file.")
    parser.add_option("-p", "--pairedEnd", dest="pairedEndFile", default=None, help="Input paired-end fastq file.")
    parser.add_option("-A", "--outputFormat", dest="outputFormat", default="sam", help="Choose an output format [sam, goby (need to re-compile with appropriate options)].")
    (options, args) = parser.parse_args()    

    #If workingDir dose not exist, should create before run the job.
    
    workingDir = os.path.dirname(options.inputFastaFile)
    
    cmds = []
    cmd_setup = "gmap_setup -d %s -D %s -k %s %s" % (options.genomeName, workingDir, options.kmer, options.inputFastaFile)
    cmds.append(cmd_setup)
    cmd_make_coords = "make -f Makefile.%s coords" % options.genomeName 
    cmds.append(cmd_make_coords)
    cmd_make_gmapdb = "make -f Makefile.%s gmapdb" % options.genomeName
    cmds.append(cmd_make_gmapdb)
    cmd_make_install = "make -f Makefile.%s install" % options.genomeName
    cmds.append(cmd_make_install)
    cmd_run = "gsnap -d %s -D %s -A %s %s " % (options.genomeName, workingDir, options.outputFormat, options.inputFastqFile)
    if options.pairedEndFile != None:
        cmd_run += "%s" % options.pairedEndFile
    cmd_run += " > %s" % options.outputfile
    cmds.append(cmd_run)
    
    tmp_files = []
    for i in range(len(cmds)):
        try:
            tmp_out = tempfile.NamedTemporaryFile().name
            tmp_files.append(tmp_out)
            tmp_stdout = open(tmp_out, 'wb')
            tmp_err = tempfile.NamedTemporaryFile().name
            tmp_files.append(tmp_err)
            tmp_stderr = open(tmp_err, 'wb')
            proc = subprocess.Popen(args=cmds[i], shell=True, cwd=".", stdout=tmp_stdout, stderr=tmp_stderr)
            returncode = proc.wait()
            tmp_stderr.close()
            #get stderr, allowing for case where it's very large
            tmp_stderr = open(tmp_err, 'rb')
            stderr = ''
            buffsize = 1048576
            try:
                while True:
                    stderr += tmp_stderr.read(buffsize)
                    if not stderr or len(stderr) % buffsize != 0:
                        break
            except OverflowError:
                pass
            tmp_stdout.close()
            tmp_stderr.close()
            if returncode != 0:
                raise Exception, stderr
        except Exception, e:
            stop_err('Error in :\n' + str(e))
    
    for tmp_file in tmp_files:
        os.remove(tmp_file)
    
if __name__=="__main__":__main__()