Mercurial > repos > lparsons > cutadapt

comparison 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
comparison
equal deleted inserted replaced
-1:000000000000 0:8b064ea16722
1 #!/usr/bin/env python
2 """
3 SYNOPSIS
4
5 cutadapt_galaxy_wrapper.py
6 -i input_file
7 -o output_file
8 [-f format (fastq/fastq/etc.)]
9 [-a 3' adapter sequence]
10 [-b 3' or 5' anywhere adapter sequence]
11 [-e error_rate]
12 [-n count]
13 [-O overlap_length]
14 [--discard discard trimmed reads]
15 [-m minimum read length]
16 [-M maximum read length]
17 [-h,--help] [-v,--verbose] [--version]
18
19 DESCRIPTION
20
21 Wrapper for cutadapt running as a galaxy tool
22
23 AUTHOR
24
25 Lance Parsons <lparsons@princeton.edu>
26
27 LICENSE
28
29 This script is in the public domain, free from copyrights or restrictions.
30
31 VERSION
32
33 $Id$
34 """
35
36 import sys, os, traceback, optparse, shutil, subprocess, tempfile
37 import re
38 #from pexpect import run, spawn
39
40 def stop_err( msg ):
41 sys.stderr.write( '%s\n' % msg )
42 sys.exit()
43
44 def main ():
45
46 global options, args
47 # Setup Parameters
48 params = []
49 if options.adapters != None:
50 params.append("-a %s" % " -a ".join(options.adapters))
51 if options.anywhere_adapters != None:
52 params.append("-b %s" % " -b ".join(options.anywhere_adapters))
53 if options.output_file != None:
54 params.append("-o %s" % options.output_file)
55 if options.error_rate != None:
56 params.append("-e %s" % options.error_rate)
57 if options.count != None:
58 params.append("-n %s" % options.count)
59 if options.overlap_length != None:
60 params.append("-O %s" % options.overlap_length)
61 if options.discard_trimmed:
62 params.append("--discard")
63 if options.minimum_length != None:
64 params.append("-m %s" % options.minimum_length)
65 if options.maximum_length != None:
66 params.append("-M %s" % options.maximum_length)
67
68 # cutadapt relies on the extension to determine file format: .fasta or .fastq
69 input_name = '.'.join((options.input,options.format))
70 # make temp directory
71 tmp_dir = tempfile.mkdtemp()
72
73 try:
74 # make a link to the input file in the tmp_dir
75 input_file = os.path.join(tmp_dir,os.path.basename(input_name))
76 os.symlink( options.input, input_file)
77
78 # generate commandline
79 cmd = 'cutadapt %s %s' % (' '.join(params),input_file)
80 proc = subprocess.Popen( args=cmd, shell=True, cwd=tmp_dir,
81 stdout=subprocess.PIPE,
82 stderr=subprocess.PIPE)
83 (stdoutdata, stderrdata) = proc.communicate()
84 returncode = proc.returncode
85 if returncode != 0:
86 raise Exception, 'Execution of cutadapt failed.\n%s' % stderrdata
87 print stderrdata
88
89 finally:
90 # clean up temp dir
91 if os.path.exists( input_name ):
92 os.remove( input_name )
93 if os.path.exists( tmp_dir ):
94 shutil.rmtree( tmp_dir )
95
96 if __name__ == '__main__':
97 try:
98 parser = optparse.OptionParser(formatter=optparse.TitledHelpFormatter(), usage=globals()['__doc__'], version='$Id$')
99 parser.add_option( '-i', '--input', dest='input', help='The sequence input file' )
100 parser.add_option( '-f', '--format', dest='format', default='fastq',
101 help='The sequence input file format (default: fastq)' )
102 parser.add_option ('-a', '--adapter', action='append', dest='adapters', help='3\' adapter sequence(s)')
103 parser.add_option ('-b', '--anywhere', action='append', dest='anywhere_adapters', help='5\' or 3\' "anywhere" adapter sequence(s)')
104 parser.add_option ('-e', '--error-rate', dest='error_rate', help='Maximum allowed error rate')
105 parser.add_option ('-n', '--times', dest='count', help='Try to remove adapters COUNT times')
106 parser.add_option ('-O', '--overlap', dest='overlap_length', help='Minimum overlap length')
107 parser.add_option ('--discard', '--discard-trimmed', dest='discard_trimmed', action='store_true', default=False, help='Discard reads that contain the adapter')
108 parser.add_option ('-m', '--minimum-length', dest='minimum_length', help='Discard reads that are shorter than LENGTH')
109 parser.add_option ('-M', '--maximum-length', dest='maximum_length', help='Discard reads that are longer than LENGTH')
110 parser.add_option ('-o', '--output', dest='output_file', help='The modified sequences are written to the file')
111 (options, args) = parser.parse_args()
112 if options.input == None:
113 stop_err("Misssing option --input")
114 if options.output_file == None:
115 stop_err("Misssing option --output")
116 if not os.path.exists(options.input):
117 stop_err("Unable to read intput file: %s" % options.input)
118 #if len(args) < 1:
119 # parser.error ('missing argument')
120 main()
121 sys.exit(0)
122 except KeyboardInterrupt, e: # Ctrl-C
123 raise e
124 except SystemExit, e: # sys.exit()
125 raise e
126 except Exception, e:
127 print 'ERROR, UNEXPECTED EXCEPTION'
128 print str(e)
129 traceback.print_exc()
130 os._exit(1)
131