PageRenderTime 89ms CodeModel.GetById 76ms app.highlight 10ms RepoModel.GetById 1ms app.codeStats 0ms

/tools/gatk/gatk_wrapper.py

https://bitbucket.org/cistrome/cistrome-harvard/
Python | 126 lines | 103 code | 15 blank | 8 comment | 22 complexity | 90f5a44f1e756b8e3020fc40fb22ed76 MD5 | raw file
  1#!/usr/bin/env python
  2#Dan Blankenberg
  3
  4"""
  5A wrapper script for running the GenomeAnalysisTK.jar commands.
  6"""
  7
  8import sys, optparse, os, tempfile, subprocess, shutil
  9from binascii import unhexlify
 10from string import Template
 11
 12GALAXY_EXT_TO_GATK_EXT = { 'gatk_interval':'intervals', 'bam_index':'bam.bai', 'gatk_dbsnp':'dbSNP', 'picard_interval_list':'interval_list' } #items not listed here will use the galaxy extension as-is
 13GALAXY_EXT_TO_GATK_FILE_TYPE = GALAXY_EXT_TO_GATK_EXT #for now, these are the same, but could be different if needed
 14DEFAULT_GATK_PREFIX = "gatk_file"
 15CHUNK_SIZE = 2**20 #1mb
 16
 17
 18def cleanup_before_exit( tmp_dir ):
 19    if tmp_dir and os.path.exists( tmp_dir ):
 20        shutil.rmtree( tmp_dir )
 21
 22def gatk_filename_from_galaxy( galaxy_filename, galaxy_ext, target_dir = None, prefix = None ):
 23    suffix = GALAXY_EXT_TO_GATK_EXT.get( galaxy_ext, galaxy_ext )
 24    if prefix is None:
 25        prefix = DEFAULT_GATK_PREFIX
 26    if target_dir is None:
 27        target_dir = os.getcwd()
 28    gatk_filename = os.path.join( target_dir, "%s.%s" % ( prefix, suffix ) )
 29    os.symlink( galaxy_filename, gatk_filename )
 30    return gatk_filename
 31
 32def gatk_filetype_argument_substitution( argument, galaxy_ext ):
 33    return argument % dict( file_type = GALAXY_EXT_TO_GATK_FILE_TYPE.get( galaxy_ext, galaxy_ext ) )
 34
 35def open_file_from_option( filename, mode = 'rb' ):
 36    if filename:
 37        return open( filename, mode = mode )
 38    return None
 39
 40def html_report_from_directory( html_out, dir ):
 41    html_out.write( '<html>\n<head>\n<title>Galaxy - GATK Output</title>\n</head>\n<body>\n<p/>\n<ul>\n' )
 42    for fname in sorted( os.listdir( dir ) ):
 43        html_out.write(  '<li><a href="%s">%s</a></li>\n' % ( fname, fname ) )
 44    html_out.write( '</ul>\n</body>\n</html>\n' )
 45
 46def index_bam_files( bam_filenames, tmp_dir ):
 47    for bam_filename in bam_filenames:
 48        bam_index_filename = "%s.bai" % bam_filename
 49        if not os.path.exists( bam_index_filename ):
 50            #need to index this bam file
 51            stderr_name = tempfile.NamedTemporaryFile( prefix = "bam_index_stderr" ).name
 52            command = 'samtools index %s %s' % ( bam_filename, bam_index_filename )
 53            proc = subprocess.Popen( args=command, shell=True, stderr=open( stderr_name, 'wb' ) )
 54            return_code = proc.wait()
 55            if return_code:
 56                for line in open( stderr_name ):
 57                    print >> sys.stderr, line
 58                os.unlink( stderr_name ) #clean up
 59                cleanup_before_exit( tmp_dir )
 60                raise Exception( "Error indexing BAM file" )
 61            os.unlink( stderr_name ) #clean up
 62
 63def __main__():
 64    #Parse Command Line
 65    parser = optparse.OptionParser()
 66    parser.add_option( '-p', '--pass_through', dest='pass_through_options', action='append', type="string", help='These options are passed through directly to GATK, without any modification.' )
 67    parser.add_option( '-o', '--pass_through_options', dest='pass_through_options_encoded', action='append', type="string", help='These options are passed through directly to GATK, with decoding from binascii.unhexlify.' )
 68    parser.add_option( '-d', '--dataset', dest='datasets', action='append', type="string", nargs=4, help='"-argument" "original_filename" "galaxy_filetype" "name_prefix"' )
 69    parser.add_option( '', '--max_jvm_heap', dest='max_jvm_heap', action='store', type="string", default=None, help='If specified, the maximum java virtual machine heap size will be set to the provide value.' )
 70    parser.add_option( '', '--max_jvm_heap_fraction', dest='max_jvm_heap_fraction', action='store', type="int", default=None, help='If specified, the maximum java virtual machine heap size will be set to the provide value as a fraction of total physical memory.' )
 71    parser.add_option( '', '--stdout', dest='stdout', action='store', type="string", default=None, help='If specified, the output of stdout will be written to this file.' )
 72    parser.add_option( '', '--stderr', dest='stderr', action='store', type="string", default=None, help='If specified, the output of stderr will be written to this file.' )
 73    parser.add_option( '', '--html_report_from_directory', dest='html_report_from_directory', action='append', type="string", nargs=2, help='"Target HTML File" "Directory"')
 74    (options, args) = parser.parse_args()
 75    
 76    tmp_dir = tempfile.mkdtemp( prefix='tmp-gatk-' )
 77    if options.pass_through_options:
 78        cmd = ' '.join( options.pass_through_options )
 79    else:
 80        cmd = ''
 81    if options.pass_through_options_encoded:
 82        cmd = '%s %s' % ( cmd, ' '.join( map( unhexlify, options.pass_through_options_encoded ) ) )
 83    if options.max_jvm_heap is not None:
 84        cmd = cmd.replace( 'java ', 'java -Xmx%s ' % ( options.max_jvm_heap ), 1 )
 85    elif options.max_jvm_heap_fraction is not None:
 86        cmd = cmd.replace( 'java ', 'java -XX:DefaultMaxRAMFraction=%s  -XX:+UseParallelGC ' % ( options.max_jvm_heap_fraction ), 1 )
 87    bam_filenames = []
 88    if options.datasets:
 89        for ( dataset_arg, filename, galaxy_ext, prefix ) in options.datasets:
 90            gatk_filename = gatk_filename_from_galaxy( filename, galaxy_ext, target_dir = tmp_dir, prefix = prefix )
 91            if dataset_arg:
 92                cmd = '%s %s "%s"' % ( cmd, gatk_filetype_argument_substitution( dataset_arg, galaxy_ext ), gatk_filename )
 93            if galaxy_ext == "bam":
 94                bam_filenames.append( gatk_filename )
 95    index_bam_files( bam_filenames, tmp_dir )
 96    #set up stdout and stderr output options
 97    stdout = open_file_from_option( options.stdout, mode = 'wb' )
 98    stderr = open_file_from_option( options.stderr, mode = 'wb' )
 99    #if no stderr file is specified, we'll use our own
100    if stderr is None:
101        stderr = tempfile.NamedTemporaryFile( prefix="gatk-stderr-", dir=tmp_dir )
102    
103    proc = subprocess.Popen( args=cmd, stdout=stdout, stderr=stderr, shell=True, cwd=tmp_dir )
104    return_code = proc.wait()
105    
106    if return_code:
107        stderr_target = sys.stderr
108    else:
109        stderr_target = sys.stdout
110    stderr.flush()
111    stderr.seek(0)
112    while True:
113        chunk = stderr.read( CHUNK_SIZE )
114        if chunk:
115            stderr_target.write( chunk )
116        else:
117            break
118    stderr.close()
119    #generate html reports
120    if options.html_report_from_directory:
121        for ( html_filename, html_dir ) in options.html_report_from_directory:
122            html_report_from_directory( open( html_filename, 'wb' ), html_dir )
123    
124    cleanup_before_exit( tmp_dir )
125
126if __name__=="__main__": __main__()