annotate spp_wrapper.py @ 3:5890ec99e05b draft default tip

Uploaded
author ziru-zhou
date Tue, 18 Dec 2012 10:45:28 -0500
parents 288ae929e6b9
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
1
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
1 #purpose: python wrapper to run spp
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
2 #author: Ziru Zhou
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
3 #Date: November 2012
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
4
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
5 import sys, subprocess, tempfile, shutil, glob, os, os.path, gzip
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
6 from galaxy import eggs
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
7 import pkg_resources
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
8 pkg_resources.require( "simplejson" )
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
9 import simplejson
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
10
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
11 CHUNK_SIZE = 1024
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
12
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
13 def main():
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
14 options = simplejson.load( open( sys.argv[1] ) )
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
15 output_narrow_peak = sys.argv[2]
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
16 output_region_peak = sys.argv[3]
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
17 output_peakshift_file = sys.argv[4]
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
18 output_rdata_file = sys.argv[5]
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
19 output_plot_file = sys.argv[6]
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
20 output_default_file = sys.argv[7]
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
21 script_path = sys.argv[8]
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
22
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
23 #set file extensions and set mandatory options
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
24 #========================================================================================
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
25 experiment_name = '_'.join( options['experiment_name'].split() ) #save experiment name
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
26
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
27 chip_file = "%s.bam" % (options['chip_file'])
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
28 subprocess.call(["cp", options['chip_file'], chip_file])
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
29
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
30 cmdline = "Rscript %s/run_spp.R -c=%s" % (script_path, chip_file )
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
31 if 'input_file' in options:
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
32 input_file = "%s.bam" % (options['input_file'])
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
33 subprocess.call(["cp", options['input_file'], input_file])
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
34 cmdline = "%s -i=%s" % ( cmdline, input_file )
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
35
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
36 #set additional options
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
37 #========================================================================================
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
38 if (options['action'] == "cross_correlation"):
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
39 cmdline = "%s %s %s %s > default_output.txt" % ( cmdline, options['savp'], options['out'], options['rf'] )
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
40 elif (options['action'] == "peak_calling"):
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
41 cmdline = "%s -fdr=%s -npeak=%s %s %s %s %s %s > default_output.txt" % ( cmdline, options['fdr'], options['npeak'], options['savr'], options['savd'], options['savn'], options['savp'], options['rf'] )
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
42 elif (options['action'] == "idr"):
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
43 cmdline = "%s -npeak=%s %s %s %s %s > default_output.txt" % ( cmdline, options['npeak'], options['savr'], options['savp'], options['out'], options['rf'] )
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
44 elif (options['action'] == "custom"):
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
45 cmdline = "%s -s=%s %s -x=%s -fdr=%s -npeak=%s %s %s" % ( cmdline, options['s'], options['speak'], options['x'], options['fdr'], options['npeak'], options['filtchr'], options['rf'] )
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
46 cmdline = "%s %s %s %s %s %s > default_output.txt" % ( cmdline, options['out'], options['savn'], options['savr'], options['savp'], options['savd'] )
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
47
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
48 #run cmdline
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
49 #========================================================================================
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
50 tmp_dir = tempfile.mkdtemp()
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
51 stderr_name = tempfile.NamedTemporaryFile().name
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
52 proc = subprocess.Popen( args=cmdline, shell=True, cwd=tmp_dir, stderr=open( stderr_name, 'wb' ) )
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
53 proc.wait()
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
54
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
55 #Do not terminate if error code, allow dataset (e.g. log) creation and cleanup
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
56 #========================================================================================
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
57 if proc.returncode:
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
58 stderr_f = open( stderr_name )
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
59 while True:
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
60 chunk = stderr_f.read( CHUNK_SIZE )
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
61 if not chunk:
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
62 stderr_f.close()
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
63 break
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
64 sys.stderr.write( chunk )
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
65
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
66
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
67 #determine if the outputs are there, copy them to the appropriate dir and filename
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
68 #========================================================================================
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
69 created_default_file = os.path.join( tmp_dir, "default_output.txt" )
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
70 if os.path.exists( created_default_file ):
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
71 shutil.move( created_default_file, output_default_file )
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
72
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
73 chip_name = os.path.basename(options['chip_file'])
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
74 input_name = os.path.basename(options['input_file'])
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
75 created_narrow_peak = os.path.join( "/mnt/galaxyData/files/000", "%s_VS_%s.narrowPeak.gz" % (chip_name, input_name) )
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
76 if os.path.exists( created_narrow_peak ):
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
77 shutil.move( created_narrow_peak, output_narrow_peak )
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
78
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
79 created_region_peak = os.path.join( "/mnt/galaxyData/files/000", "%s_VS_%s.regionPeak.gz" % (chip_name, input_name) )
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
80 if os.path.exists( created_region_peak ):
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
81 shutil.move( created_region_peak, output_region_peak )
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
82
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
83 created_peakshift_file = os.path.join( tmp_dir, "peakshift.txt" )
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
84 if os.path.exists( created_peakshift_file ):
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
85 shutil.move( created_peakshift_file, output_peakshift_file )
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
86
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
87 created_rdata_file = os.path.join( tmp_dir, "%s.Rdata" % options['chip_file'] )
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
88 if os.path.exists( created_rdata_file ):
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
89 shutil.move( created_rdata_file, output_rdata_file )
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
90
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
91 created_plot_file = os.path.join( tmp_dir, "%s.pdf" % options['chip_file'] )
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
92 if os.path.exists( created_plot_file ):
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
93 shutil.move( created_plot_file, output_plot_file )
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
94
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
95
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
96 os.unlink( stderr_name )
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
97 #os.rmdir( tmp_dir )
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
98
288ae929e6b9 Uploaded
ziru-zhou
parents:
diff changeset
99 if __name__ == "__main__": main()