comparison spp_wrapper.py @ 7:cbbc26d6ca30 draft

Uploaded
author modencode-dcc
date Fri, 18 Jan 2013 18:48:21 -0500
parents
children 2db63ab9b129
comparison
equal deleted inserted replaced
6:a4f0a7862da6 7:cbbc26d6ca30
1 #purpose: python wrapper to run spp
2 #author: Ziru Zhou
3 #Date: November 2012
4
5 import sys, subprocess, tempfile, shutil, glob, os, os.path, gzip
6 from galaxy import eggs
7 import pkg_resources
8 pkg_resources.require( "simplejson" )
9 import simplejson
10
11 CHUNK_SIZE = 1024
12
13 def main():
14 options = simplejson.load( open( sys.argv[1] ) )
15 output_narrow_peak = sys.argv[2]
16 output_region_peak = sys.argv[3]
17 output_peakshift_file = sys.argv[4]
18 output_rdata_file = sys.argv[5]
19 output_plot_file = sys.argv[6]
20 output_default_file = sys.argv[7]
21 script_path = sys.argv[8]
22
23 #set file extensions and set mandatory options
24 #========================================================================================
25 experiment_name = '_'.join( options['experiment_name'].split() ) #save experiment name
26
27 chip_file = "%s.bam" % (options['chip_file'])
28 subprocess.call(["cp", options['chip_file'], chip_file])
29
30 cmdline = "Rscript %s/run_spp.R -c=%s" % (script_path, chip_file )
31 if 'input_file' in options:
32 input_file = "%s.bam" % (options['input_file'])
33 subprocess.call(["cp", options['input_file'], input_file])
34 cmdline = "%s -i=%s" % ( cmdline, input_file )
35
36 #set additional options
37 #========================================================================================
38 if (options['action'] == "cross_correlation"):
39 cmdline = "%s %s %s %s > default_output.txt" % ( cmdline, options['savp'], options['out'], options['rf'] )
40 elif (options['action'] == "peak_calling"):
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'] )
42 elif (options['action'] == "idr"):
43 cmdline = "%s -npeak=%s %s %s %s %s > default_output.txt" % ( cmdline, options['npeak'], options['savr'], options['savp'], options['out'], options['rf'] )
44 elif (options['action'] == "custom"):
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'] )
46 cmdline = "%s %s %s %s %s %s > default_output.txt" % ( cmdline, options['out'], options['savn'], options['savr'], options['savp'], options['savd'] )
47
48 #run cmdline
49 #========================================================================================
50 tmp_dir = tempfile.mkdtemp()
51 stderr_name = tempfile.NamedTemporaryFile().name
52 proc = subprocess.Popen( args=cmdline, shell=True, cwd=tmp_dir, stderr=open( stderr_name, 'wb' ) )
53 proc.wait()
54
55 #Do not terminate if error code, allow dataset (e.g. log) creation and cleanup
56 #========================================================================================
57 if proc.returncode:
58 stderr_f = open( stderr_name )
59 while True:
60 chunk = stderr_f.read( CHUNK_SIZE )
61 if not chunk:
62 stderr_f.close()
63 break
64 sys.stderr.write( chunk )
65
66
67 #determine if the outputs are there, copy them to the appropriate dir and filename
68 #========================================================================================
69 created_default_file = os.path.join( tmp_dir, "default_output.txt" )
70 if os.path.exists( created_default_file ):
71 shutil.move( created_default_file, output_default_file )
72
73 chip_name = os.path.basename(options['chip_file'])
74 input_name = os.path.basename(options['input_file'])
75 created_narrow_peak = os.path.join( "/mnt/galaxyData/files/000", "%s_VS_%s.narrowPeak.gz" % (chip_name, input_name) )
76 if os.path.exists( created_narrow_peak ):
77 shutil.move( created_narrow_peak, output_narrow_peak )
78
79 created_region_peak = os.path.join( "/mnt/galaxyData/files/000", "%s_VS_%s.regionPeak.gz" % (chip_name, input_name) )
80 if os.path.exists( created_region_peak ):
81 shutil.move( created_region_peak, output_region_peak )
82
83 created_peakshift_file = os.path.join( tmp_dir, "peakshift.txt" )
84 if os.path.exists( created_peakshift_file ):
85 shutil.move( created_peakshift_file, output_peakshift_file )
86
87 created_rdata_file = os.path.join( tmp_dir, "%s.Rdata" % options['chip_file'] )
88 if os.path.exists( created_rdata_file ):
89 shutil.move( created_rdata_file, output_rdata_file )
90
91 created_plot_file = os.path.join( tmp_dir, "%s.pdf" % options['chip_file'] )
92 if os.path.exists( created_plot_file ):
93 shutil.move( created_plot_file, output_plot_file )
94
95
96 os.unlink( stderr_name )
97 #os.rmdir( tmp_dir )
98
99 if __name__ == "__main__": main()