|
0
|
1 #!/usr/bin/env python
|
|
|
2 #Dan Blankenberg
|
|
|
3
|
|
|
4 """
|
|
|
5 A wrapper script for running the Picard SamToFastq command. Allows parsing read groups into separate files.
|
|
|
6 """
|
|
|
7
|
|
|
8 import sys, optparse, os, tempfile, subprocess, shutil
|
|
|
9
|
|
|
10 CHUNK_SIZE = 2**20 #1mb
|
|
|
11
|
|
|
12
|
|
|
13 def cleanup_before_exit( tmp_dir ):
|
|
|
14 if tmp_dir and os.path.exists( tmp_dir ):
|
|
|
15 shutil.rmtree( tmp_dir )
|
|
|
16
|
|
|
17 def open_file_from_option( filename, mode = 'rb' ):
|
|
|
18 if filename:
|
|
|
19 return open( filename, mode = mode )
|
|
|
20 return None
|
|
|
21
|
|
|
22 def __main__():
|
|
|
23 #Parse Command Line
|
|
|
24 parser = optparse.OptionParser()
|
|
|
25 parser.add_option( '-p', '--pass_through', dest='pass_through_options', action='append', type="string", help='These options are passed through directly to PICARD, without any modification.' )
|
|
|
26 parser.add_option( '-1', '--read_group_file_1', dest='read_group_file_1', action='store', type="string", default=None, help='Read Group 1 output file, when using multiple readgroups' )
|
|
|
27 parser.add_option( '-2', '--read_group_file_2', dest='read_group_file_2', action='store', type="string", default=None, help='Read Group 2 output file, when using multiple readgroups and paired end' )
|
|
|
28 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.' )
|
|
|
29 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.' )
|
|
|
30 parser.add_option( '-n', '--new_files_path', dest='new_files_path', action='store', type="string", default=None, help='new_files_path')
|
|
|
31 parser.add_option( '-i', '--file_id_1', dest='file_id_1', action='store', type="string", default=None, help='file_id_1')
|
|
|
32 parser.add_option( '-f', '--file_id_2', dest='file_id_2', action='store', type="string", default=None, help='file_id_2')
|
|
|
33 (options, args) = parser.parse_args()
|
|
|
34
|
|
|
35 tmp_dir = tempfile.mkdtemp( prefix='tmp-picard-' )
|
|
|
36 if options.pass_through_options:
|
|
|
37 cmd = ' '.join( options.pass_through_options )
|
|
|
38 else:
|
|
|
39 cmd = ''
|
|
|
40 if options.new_files_path is not None:
|
|
|
41 print 'Creating FASTQ files by Read Group'
|
|
|
42 assert None not in [ options.read_group_file_1, options.new_files_path, options.file_id_1 ], 'When using read group aware, you need to specify --read_group_file_1, --read_group_file_2 (when paired end), --new_files_path, and --file_id'
|
|
|
43 cmd = '%s OUTPUT_DIR="%s"' % ( cmd, tmp_dir)
|
|
|
44 #set up stdout and stderr output options
|
|
|
45 stdout = open_file_from_option( options.stdout, mode = 'wb' )
|
|
|
46 if stdout is None:
|
|
|
47 stdout = sys.stdout
|
|
|
48 stderr = open_file_from_option( options.stderr, mode = 'wb' )
|
|
|
49 #if no stderr file is specified, we'll use our own
|
|
|
50 if stderr is None:
|
|
|
51 stderr = tempfile.NamedTemporaryFile( prefix="picard-stderr-", dir=tmp_dir )
|
|
|
52
|
|
|
53 proc = subprocess.Popen( args=cmd, stdout=stdout, stderr=stderr, shell=True, cwd=tmp_dir )
|
|
|
54 return_code = proc.wait()
|
|
|
55
|
|
|
56 if return_code:
|
|
|
57 stderr_target = sys.stderr
|
|
|
58 else:
|
|
|
59 stderr_target = sys.stdout
|
|
|
60 stderr.flush()
|
|
|
61 stderr.seek(0)
|
|
|
62 while True:
|
|
|
63 chunk = stderr.read( CHUNK_SIZE )
|
|
|
64 if chunk:
|
|
|
65 stderr_target.write( chunk )
|
|
|
66 else:
|
|
|
67 break
|
|
|
68 stderr.close()
|
|
|
69 #if rg aware, put files where they belong
|
|
|
70 if options.new_files_path is not None:
|
|
|
71 fastq_1_name = options.read_group_file_1
|
|
|
72 fastq_2_name = options.read_group_file_2
|
|
|
73 file_id_1 = options.file_id_1
|
|
|
74 file_id_2 = options.file_id_2
|
|
|
75 if file_id_2 is None:
|
|
|
76 file_id_2 = file_id_1
|
|
|
77 for filename in sorted( os.listdir( tmp_dir ) ):
|
|
|
78 if filename.endswith( '_1.fastq' ):
|
|
|
79 if fastq_1_name:
|
|
|
80 shutil.move( os.path.join( tmp_dir, filename ), fastq_1_name )
|
|
|
81 fastq_1_name = None
|
|
|
82 else:
|
|
|
83 shutil.move( os.path.join( tmp_dir, filename ), os.path.join( options.new_files_path, 'primary_%s_%s - 1_visible_fastqsanger' % ( file_id_1, filename[:-len( '_1.fastq' )] ) ) )
|
|
|
84 elif filename.endswith( '_2.fastq' ):
|
|
|
85 if fastq_2_name:
|
|
|
86 shutil.move( os.path.join( tmp_dir, filename ), fastq_2_name )
|
|
|
87 fastq_2_name = None
|
|
|
88 else:
|
|
|
89 shutil.move( os.path.join( tmp_dir, filename ), os.path.join( options.new_files_path, 'primary_%s_%s - 2_visible_fastqsanger' % ( file_id_2, filename[:-len( '_2.fastq' )] ) ) )
|
|
|
90
|
|
|
91 cleanup_before_exit( tmp_dir )
|
|
|
92
|
|
|
93 if __name__=="__main__": __main__()
|
