annotate oases_optimiser.py @ 4:42fad5f9c78e draft

planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 4ea89b80296509604d1b224ee1671a6497ddca14
author drosofff
date Mon, 19 Oct 2015 11:56:34 -0400
parents b4d1940ecc90
children 58bf34f98232
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
1 #!/usr/bin/env python
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
2
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
3 """
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
4 VelvetOptimiser Wrapper
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
5 refactored using the adaptation of
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
6 Konrad Paszkiewicz University of Exeter, UK.
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
7
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
8 """
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
9 import pkg_resources;
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
10 import logging, os, string, sys, tempfile, glob, shutil, types, urllib
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
11 import shlex, subprocess
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
12 from optparse import OptionParser, OptionGroup
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
13 from stat import *
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
14
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
15
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
16 def stop_err( msg ):
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
17 sys.stderr.write( "%s\n" % msg )
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
18 sys.exit()
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
19
3
b4d1940ecc90 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents: 0
diff changeset
20 def oases_optimiser(starthash, endhash, input, job_dir):
b4d1940ecc90 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents: 0
diff changeset
21 '''
b4d1940ecc90 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents: 0
diff changeset
22 Replaces call to oases_optimiser.sh. For all k-mers between
b4d1940ecc90 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents: 0
diff changeset
23 starthash and endhash run velvet and oases.
b4d1940ecc90 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents: 0
diff changeset
24 '''
b4d1940ecc90 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents: 0
diff changeset
25 for i in xrange(starthash, endhash, 2):
b4d1940ecc90 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents: 0
diff changeset
26 cmd1="velveth {0}/outputFolder_{1} {1} {2} && ".format(job_dir, i, input)
b4d1940ecc90 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents: 0
diff changeset
27 cmd2="velvetg {0}/outputFolder_{1} -read_trkg yes && ".format(job_dir, i)
b4d1940ecc90 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents: 0
diff changeset
28 cmd3="oases {0}/outputFolder_{1}".format(job_dir, i)
b4d1940ecc90 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents: 0
diff changeset
29 proc = subprocess.call( args=cmd1+cmd2+cmd3, shell=True, stdout=sys.stdout, stderr=sys.stderr )
b4d1940ecc90 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents: 0
diff changeset
30 cmd4="velveth {0}/MergedAssemblyFolder 27 -long outputFolder_*/transcripts.fa && ".format(job_dir)
b4d1940ecc90 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents: 0
diff changeset
31 cmd5="velvetg {0}/MergedAssemblyFolder -read_trkg yes -conserveLong yes && ".format(job_dir)
b4d1940ecc90 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents: 0
diff changeset
32 cmd6="oases {0}/MergedAssemblyFolder -merge yes".format(job_dir)
b4d1940ecc90 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents: 0
diff changeset
33 proc = subprocess.call( args=cmd4+cmd5+cmd6, shell=True, stdout=sys.stdout, stderr=sys.stderr )
b4d1940ecc90 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents: 0
diff changeset
34
0
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
35 def __main__():
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
36 job_dir= os.getcwd()
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
37 #Parse Command Line
3
b4d1940ecc90 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents: 0
diff changeset
38 starthash = int(sys.argv[1])
b4d1940ecc90 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents: 0
diff changeset
39 endhash = int(sys.argv[2])
b4d1940ecc90 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents: 0
diff changeset
40 input = sys.argv[3]
0
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
41 transcripts = sys.argv[4]
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
42 transcripts_path = ''
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
43 print >> sys.stdout, "PATH = %s" % (os.environ['PATH'])
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
44 try:
3
b4d1940ecc90 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents: 0
diff changeset
45 oases_optimiser(starthash, endhash, input, job_dir)
0
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
46 except Exception, e:
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
47 stop_err( 'Error running oases_optimiser.py' + str( e ) )
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
48 out = open(transcripts,'w')
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
49 transcript_path = os.path.join(job_dir, "MergedAssemblyFolder", 'transcripts.fa')
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
50 print >> sys.stdout, transcript_path
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
51 for line in open(transcript_path):
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
52 out.write( "%s" % (line) )
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
53 out.close()
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
54
435711ba68e3 planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
55 if __name__ == "__main__": __main__()