annotate samtools_wrapper.py @ 1:dddde948c3e5

Update samtools_mpileup to use the fasta_indexes data table.
author Dave Bouvier <dave@bx.psu.edu>
date Wed, 11 Dec 2013 12:54:32 -0500
parents 44a18a94d7a9
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
1 #!/usr/bin/env python
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
2 #Dan Blankenberg
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
3
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
4 """
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
5 A wrapper script for running SAMTools commands.
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
6 """
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
7
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
8 import sys, optparse, os, tempfile, subprocess, shutil
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
9 from string import Template
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
10
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
11 GALAXY_EXT_TO_SAMTOOLS_EXT = { 'bam_index':'bam.bai', } #items not listed here will use the galaxy extension as-is
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
12 GALAXY_EXT_TO_SAMTOOLS_FILE_TYPE = GALAXY_EXT_TO_SAMTOOLS_EXT #for now, these are the same, but could be different if needed
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
13 DEFAULT_SAMTOOLS_PREFIX = "SAMTools_file"
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
14 CHUNK_SIZE = 2**20 #1mb
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
15
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
16
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
17 def cleanup_before_exit( tmp_dir ):
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
18 if tmp_dir and os.path.exists( tmp_dir ):
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
19 shutil.rmtree( tmp_dir )
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
20
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
21 def SAMTOOLS_filename_from_galaxy( galaxy_filename, galaxy_ext, target_dir = None, prefix = None ):
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
22 suffix = GALAXY_EXT_TO_SAMTOOLS_EXT.get( galaxy_ext, galaxy_ext )
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
23 if prefix is None:
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
24 prefix = DEFAULT_SAMTOOLS_PREFIX
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
25 if target_dir is None:
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
26 target_dir = os.getcwd()
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
27 SAMTools_filename = os.path.join( target_dir, "%s.%s" % ( prefix, suffix ) )
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
28 os.symlink( galaxy_filename, SAMTools_filename )
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
29 return SAMTools_filename
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
30
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
31 def SAMTOOLS_filetype_argument_substitution( argument, galaxy_ext ):
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
32 return argument % dict( file_type = GALAXY_EXT_TO_SAMTOOLS_FILE_TYPE.get( galaxy_ext, galaxy_ext ) )
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
33
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
34 def open_file_from_option( filename, mode = 'rb' ):
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
35 if filename:
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
36 return open( filename, mode = mode )
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
37 return None
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
38
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
39 def html_report_from_directory( html_out, dir ):
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
40 html_out.write( '<html>\n<head>\n<title>Galaxy - SAMTOOLS Output</title>\n</head>\n<body>\n<p/>\n<ul>\n' )
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
41 for fname in sorted( os.listdir( dir ) ):
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
42 html_out.write( '<li><a href="%s">%s</a></li>\n' % ( fname, fname ) )
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
43 html_out.write( '</ul>\n</body>\n</html>\n' )
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
44
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
45 def __main__():
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
46 #Parse Command Line
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
47 parser = optparse.OptionParser()
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
48 parser.add_option( '-p', '--pass_through', dest='pass_through_options', action='append', type="string", help='These options are passed through directly to SAMTOOLS, without any modification.' )
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
49 parser.add_option( '-d', '--dataset', dest='datasets', action='append', type="string", nargs=4, help='"-argument" "original_filename" "galaxy_filetype" "name_prefix"' )
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
50 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.' )
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
51 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.' )
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
52 parser.add_option( '', '--html_report_from_directory', dest='html_report_from_directory', action='append', type="string", nargs=2, help='"Target HTML File" "Directory"')
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
53 (options, args) = parser.parse_args()
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
54
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
55 tmp_dir = tempfile.mkdtemp( prefix='tmp-SAMTOOLS-' )
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
56
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
57 #set up stdout and stderr output options
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
58 stdout = open_file_from_option( options.stdout, mode = 'wb' )
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
59 stderr = open_file_from_option( options.stderr, mode = 'wb' )
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
60 #if no stderr file is specified, we'll use our own
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
61 if stderr is None:
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
62 stderr = tempfile.NamedTemporaryFile( prefix="SAMTOOLS-stderr-", dir=tmp_dir )
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
63
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
64 if options.pass_through_options:
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
65 cmd = ' '.join( options.pass_through_options )
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
66 else:
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
67 cmd = ''
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
68 return_code = None
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
69 if options.datasets:
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
70 for ( dataset_arg, filename, galaxy_ext, prefix ) in options.datasets:
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
71 SAMTools_filename = SAMTOOLS_filename_from_galaxy( filename, galaxy_ext, target_dir = tmp_dir, prefix = prefix )
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
72 if dataset_arg:
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
73 if '>' in cmd:
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
74 cmd = cmd.replace( '>', ' %s "%s" >' % ( SAMTOOLS_filetype_argument_substitution( dataset_arg, galaxy_ext ), SAMTools_filename ), 1 )
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
75 else:
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
76 cmd = '%s %s "%s"' % ( cmd, SAMTOOLS_filetype_argument_substitution( dataset_arg, galaxy_ext ), SAMTools_filename )
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
77 #auto index fasta files:
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
78 if galaxy_ext == 'fa':
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
79 index_cmd = 'samtools faidx %s' % ( SAMTools_filename )
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
80 proc = subprocess.Popen( args=index_cmd, stdout=stdout, stderr=stderr, shell=True, cwd=tmp_dir )
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
81 return_code = proc.wait()
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
82 if return_code:
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
83 break
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
84 if return_code is None or not return_code:
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
85 proc = subprocess.Popen( args=cmd, stdout=stdout, stderr=stderr, shell=True, cwd=tmp_dir )
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
86 return_code = proc.wait()
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
87 if return_code:
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
88 stderr_target = sys.stderr
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
89 else:
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
90 if stdout:
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
91 stderr_target = stdout
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
92 else:
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
93 stderr_target = sys.stdout
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
94 stderr.flush()
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
95 stderr.seek(0)
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
96 while True:
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
97 chunk = stderr.read( CHUNK_SIZE )
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
98 if chunk:
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
99 stderr_target.write( chunk )
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
100 else:
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
101 break
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
102 stderr.close()
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
103 #generate html reports
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
104 if options.html_report_from_directory:
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
105 for ( html_filename, html_dir ) in options.html_report_from_directory:
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
106 html_report_from_directory( open( html_filename, 'wb' ), html_dir )
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
107
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
108 cleanup_before_exit( tmp_dir )
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
109
44a18a94d7a9 Uploaded tool tarball.
devteam
parents:
diff changeset
110 if __name__=="__main__": __main__()