Mercurial > repos > lparsons > cutadapt

annotate cutadapt_galaxy_wrapper.py @ 0:8b064ea16722

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Initial version with multiple adapter support
author Lance Parsons <lparsons@princeton.edu>
date Fri, 13 May 2011 15:54:01 -0400
parents
children 7ed26fc9fa8a
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
1 #!/usr/bin/env python
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
2 """
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
3 SYNOPSIS
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
4
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
5 cutadapt_galaxy_wrapper.py
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
6 -i input_file
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
7 -o output_file
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
8 [-f format (fastq/fastq/etc.)]
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
9 [-a 3' adapter sequence]
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
10 [-b 3' or 5' anywhere adapter sequence]
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
11 [-e error_rate]
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
12 [-n count]
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
13 [-O overlap_length]
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
14 [--discard discard trimmed reads]
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
15 [-m minimum read length]
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
16 [-M maximum read length]
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
17 [-h,--help] [-v,--verbose] [--version]
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
18
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
19 DESCRIPTION
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
20
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
21 Wrapper for cutadapt running as a galaxy tool
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
22
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
23 AUTHOR
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
24
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
25 Lance Parsons <lparsons@princeton.edu>
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
26
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
27 LICENSE
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
28
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
29 This script is in the public domain, free from copyrights or restrictions.
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
30
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
31 VERSION
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
32
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
33 $Id$
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
34 """
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
35
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
36 import sys, os, traceback, optparse, shutil, subprocess, tempfile
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
37 import re
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
38 #from pexpect import run, spawn
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
39
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
40 def stop_err( msg ):
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
41 sys.stderr.write( '%s\n' % msg )
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
42 sys.exit()
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
43
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
44 def main ():
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
45
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
46 global options, args
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
47 # Setup Parameters
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
48 params = []
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
49 if options.adapters != None:
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
50 params.append("-a %s" % " -a ".join(options.adapters))
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
51 if options.anywhere_adapters != None:
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
52 params.append("-b %s" % " -b ".join(options.anywhere_adapters))
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
53 if options.output_file != None:
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
54 params.append("-o %s" % options.output_file)
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
55 if options.error_rate != None:
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
56 params.append("-e %s" % options.error_rate)
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
57 if options.count != None:
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
58 params.append("-n %s" % options.count)
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
59 if options.overlap_length != None:
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
60 params.append("-O %s" % options.overlap_length)
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
61 if options.discard_trimmed:
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
62 params.append("--discard")
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
63 if options.minimum_length != None:
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
64 params.append("-m %s" % options.minimum_length)
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
65 if options.maximum_length != None:
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
66 params.append("-M %s" % options.maximum_length)
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
67
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
68 # cutadapt relies on the extension to determine file format: .fasta or .fastq
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
69 input_name = '.'.join((options.input,options.format))
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
70 # make temp directory
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
71 tmp_dir = tempfile.mkdtemp()
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
72
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
73 try:
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
74 # make a link to the input file in the tmp_dir
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
75 input_file = os.path.join(tmp_dir,os.path.basename(input_name))
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
76 os.symlink( options.input, input_file)
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
77
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
78 # generate commandline
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
79 cmd = 'cutadapt %s %s' % (' '.join(params),input_file)
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
80 proc = subprocess.Popen( args=cmd, shell=True, cwd=tmp_dir,
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
81 stdout=subprocess.PIPE,
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
82 stderr=subprocess.PIPE)
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
83 (stdoutdata, stderrdata) = proc.communicate()
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
84 returncode = proc.returncode
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
85 if returncode != 0:
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
86 raise Exception, 'Execution of cutadapt failed.\n%s' % stderrdata
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
87 print stderrdata
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
88
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
89 finally:
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
90 # clean up temp dir
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
91 if os.path.exists( input_name ):
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
92 os.remove( input_name )
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
93 if os.path.exists( tmp_dir ):
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
94 shutil.rmtree( tmp_dir )
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
95
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
96 if __name__ == '__main__':
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
97 try:
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
98 parser = optparse.OptionParser(formatter=optparse.TitledHelpFormatter(), usage=globals()['__doc__'], version='$Id$')
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
99 parser.add_option( '-i', '--input', dest='input', help='The sequence input file' )
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
100 parser.add_option( '-f', '--format', dest='format', default='fastq',
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
101 help='The sequence input file format (default: fastq)' )
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
102 parser.add_option ('-a', '--adapter', action='append', dest='adapters', help='3\' adapter sequence(s)')
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
103 parser.add_option ('-b', '--anywhere', action='append', dest='anywhere_adapters', help='5\' or 3\' "anywhere" adapter sequence(s)')
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
104 parser.add_option ('-e', '--error-rate', dest='error_rate', help='Maximum allowed error rate')
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
105 parser.add_option ('-n', '--times', dest='count', help='Try to remove adapters COUNT times')
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
106 parser.add_option ('-O', '--overlap', dest='overlap_length', help='Minimum overlap length')
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
107 parser.add_option ('--discard', '--discard-trimmed', dest='discard_trimmed', action='store_true', default=False, help='Discard reads that contain the adapter')
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
108 parser.add_option ('-m', '--minimum-length', dest='minimum_length', help='Discard reads that are shorter than LENGTH')
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
109 parser.add_option ('-M', '--maximum-length', dest='maximum_length', help='Discard reads that are longer than LENGTH')
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
110 parser.add_option ('-o', '--output', dest='output_file', help='The modified sequences are written to the file')
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
111 (options, args) = parser.parse_args()
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
112 if options.input == None:
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
113 stop_err("Misssing option --input")
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
114 if options.output_file == None:
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
115 stop_err("Misssing option --output")
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
116 if not os.path.exists(options.input):
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
117 stop_err("Unable to read intput file: %s" % options.input)
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
118 #if len(args) < 1:
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
119 # parser.error ('missing argument')
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
120 main()
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
121 sys.exit(0)
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
122 except KeyboardInterrupt, e: # Ctrl-C
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
123 raise e
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
124 except SystemExit, e: # sys.exit()
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
125 raise e
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
126 except Exception, e:
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
127 print 'ERROR, UNEXPECTED EXCEPTION'
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
128 print str(e)
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
129 traceback.print_exc()
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
130 os._exit(1)
8b064ea16722 Initial version with multiple adapter support
Lance Parsons <lparsons@princeton.edu>
parents:
diff changeset
131