Mercurial > repos > konradpaszkiewicz > interproscan

annotate interproscan.py @ 0:49e20fa2c66d

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
author konradpaszkiewicz
date Tue, 07 Jun 2011 17:27:16 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
1 #!/usr/bin/env python
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
2
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
3 """
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
4 Classes encapsulating decypher tool.
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
5 James E Johnson - University of Minnesota
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
6 """
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
7 import pkg_resources;
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
8 import logging, os, string, sys, tempfile, glob, shutil, types, urllib
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
9 import shlex, subprocess
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
10 from optparse import OptionParser, OptionGroup
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
11 from stat import *
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
12
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
13
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
14 log = logging.getLogger( __name__ )
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
15
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
16 assert sys.version_info[:2] >= ( 2, 4 )
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
17
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
18 def stop_err( msg ):
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
19 sys.stderr.write( "%s\n" % msg )
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
20 sys.exit()
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
21
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
22 def __main__():
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
23 #Parse Command Line
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
24 s = 'interproscan.py: argv = %s\n' % (sys.argv)
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
25 # print >> sys.stderr, s # so will appear as blurb for file
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
26 argcnt = len(sys.argv)
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
27 working_dir = sys.argv[1]
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
28 input = sys.argv[2]
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
29 format = sys.argv[3]
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
30 output = sys.argv[4]
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
31 #Convert all spaces in ORF header to underscores
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
32 cmdline = 'sed \'s/ /_/\' %s > temp.fa' % (input)
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
33 #print >> sys.stderr, cmdline
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
34 try:
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
35 proc = subprocess.Popen( args=cmdline, shell=True, stderr=subprocess.PIPE )
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
36 returncode = proc.wait()
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
37 # get stderr, allowing for case where it's very large
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
38 stderr = ''
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
39 buffsize = 1048576
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
40 try:
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
41 while True:
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
42 stderr += proc.stderr.read( buffsize )
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
43 if not stderr or len( stderr ) % buffsize != 0:
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
44 break
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
45 except OverflowError:
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
46 pass
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
47 if returncode != 0:
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
48 raise Exception, stderr
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
49 except Exception, e:
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
50 stop_err( 'Error running sed ' + str( e ) )
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
51
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
52 cmdline = 'iprscan -cli -nocrc -i temp.fa -o temp.iprscan -goterms -seqtype p -altjobs -format %s -appl hmmpfam > /dev/null' % (format)
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
53 #print >> sys.stderr, cmdline # so will appear as blurb for file
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
54 try:
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
55 proc = subprocess.Popen( args=cmdline, shell=True, stderr=subprocess.PIPE )
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
56 returncode = proc.wait()
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
57 # get stderr, allowing for case where it's very large
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
58 stderr = ''
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
59 buffsize = 1048576
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
60 try:
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
61 while True:
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
62 stderr += proc.stderr.read( buffsize )
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
63 if not stderr or len( stderr ) % buffsize != 0:
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
64 break
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
65 except OverflowError:
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
66 pass
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
67 if returncode != 0:
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
68 raise Exception, stderr
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
69 except Exception, e:
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
70 stop_err( 'Error running iprscan ' + str( e ) )
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
71
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
72 out = open(output,'w')
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
73 #outpe_path = os.path.join(working_dir,'')
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
74 for line in open('temp.iprscan'):
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
75 out.write( "%s" % (line) )
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
76 out.close()
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
77
49e20fa2c66d Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
konradpaszkiewicz
parents:
diff changeset
78 if __name__ == "__main__": __main__()