annotate strelka_wrapper.py @ 16:024468e8638c

environment variable
author mini
date Wed, 01 Oct 2014 12:33:46 +0200
parents ca84a74ff567
children 8260a4188b08
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
6
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
1 #!/usr/bin/env python
10
f66f23fcc83a Modified layout
mini
parents: 8
diff changeset
2 #Gregoire Seguin-Henry (Engineer IT)
f66f23fcc83a Modified layout
mini
parents: 8
diff changeset
3 #Amine Sbitti (Data Scientist)
f66f23fcc83a Modified layout
mini
parents: 8
diff changeset
4 #Ludovic Marie-Sainte (Project Manager)
f66f23fcc83a Modified layout
mini
parents: 8
diff changeset
5 #For Geviteam 2014
6
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
6
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
7 """
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
8 A wrapper script for running the GenomeAnalysisTK.jar commands.
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
9 """
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
10 from __future__ import print_function
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
11 import sys, argparse, os, tempfile, subprocess, shutil
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
12 from binascii import unhexlify
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
13 from string import Template
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
14 from galaxy import eggs
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
15
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
16 def cleanup_before_exit( tmp_dir ):
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
17 if tmp_dir and os.path.exists( tmp_dir ):
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
18 shutil.rmtree( tmp_dir )
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
19
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
20 def _create_config(args, config_path):
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
21 conf_file = open(config_path, "w")
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
22 conf_file.write("[user]\n")
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
23 for option in args:
7
3e8541ece3c7 trying to fix environment variable 0.0.1
mini
parents: 6
diff changeset
24 if not option in ["tumorBam", "normalBam", "refFile", "configFile", "scriptPath"] and args[option]!=None:
6
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
25 conf_file.write("%s=%s\n" % (option, args[option]))
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
26 conf_file.close()
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
27
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
28 def my_Popen(cmd, prefix_for_stderr_name, tmp_dir, msg_error):
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
29 stderr_name = tempfile.NamedTemporaryFile( prefix = prefix_for_stderr_name ).name
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
30 proc = subprocess.Popen( args=cmd, shell=True, stderr=open( stderr_name, 'wb' ) )
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
31 return_code = proc.wait()
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
32 if return_code:
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
33 for line in open( stderr_name ):
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
34 print(line, file=sys.stderr)
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
35 os.unlink( stderr_name ) #clean up
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
36 cleanup_before_exit( tmp_dir )
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
37 raise Exception( msg_error )
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
38 else:
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
39 os.unlink( stderr_name )
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
40
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
41 def index_bam_files( bam_filenames, tmp_dir ):
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
42 for bam_filename in bam_filenames:
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
43 bam_index_filename = "%s.bai" % bam_filename
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
44 print("bam_filename is: " + bam_filename + " bam_index_filename is: " + bam_index_filename + " test is: %s" % os.path.exists(bam_index_filename))
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
45 if not os.path.exists( bam_index_filename ):
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
46 #need to index this bam file
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
47 command = 'samtools index %s %s' % ( bam_filename, bam_index_filename )
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
48 my_Popen( command, "bam_index_stderr", tmp_dir, "Error during indexation of fasta file :" + bam_filename)
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
49
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
50 def index_fasta_files( fasta_filenames, tmp_dir ):
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
51 for fasta_filename in fasta_filenames:
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
52 fasta_index_filename = "%s.fai" % fasta_filename
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
53 print("fasta_filename is: " + fasta_filename + " fasta_index_filename is: " + fasta_index_filename + " test is: %s" % os.path.exists(fasta_index_filename))
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
54 if not os.path.exists( fasta_index_filename ):
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
55 #need to index this bam file
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
56 command = 'samtools faidx %s %s' % ( fasta_filename, fasta_index_filename )
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
57 my_Popen( command, "fasta_index_stderr", tmp_dir, "Error during indexation of fasta file :" + fasta_filename)
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
58
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
59 def __main__():
10
f66f23fcc83a Modified layout
mini
parents: 8
diff changeset
60 #Manage options
6
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
61 parser = argparse.ArgumentParser()
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
62 parser.add_argument( '-t', '--tumorBam', help='path to tumor bam file', required = False )
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
63 parser.add_argument( '-n', '--normalBam', help='path to tumor bam file', required = False )
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
64 parser.add_argument( '-r', '--refFile', help='path to tumor bam file', required = False )
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
65 parser.add_argument( '-c', '--configFile', help='path to tumor bam file', required = False )
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
66 parser.add_argument( '--depthFilterMultiple', help='path to tumor bam file', required = False )
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
67 parser.add_argument( '--snvMaxFilteredBasecallFrac', help='path to tumor bam file', required = False )
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
68 parser.add_argument( '--snvMaxSpanningDeletionFrac', help='path to tumor bam file', required = False )
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
69 parser.add_argument( '--indelMaxRefRepeat', help='path to tumor bam file', required = False )
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
70 parser.add_argument( '--indelMaxWindowFilteredBasecallFrac', help='path to tumor bam file', required = False )
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
71 parser.add_argument( '--indelMaxIntHpolLength', help='path to tumor bam file', required = False )
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
72 parser.add_argument( '--ssnvPrior', help='path to tumor bam file', required = False )
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
73 parser.add_argument( '--sindelPrior', help='path to tumor bam file', required = False )
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
74 parser.add_argument( '--ssnvNoise', help='path to tumor bam file', required = False )
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
75 parser.add_argument( '--sindelNoise', help='path to tumor bam file', required = False )
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
76 parser.add_argument( '--ssnvNoiseStrandBiasFrac', help='path to tumor bam file', required = False )
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
77 parser.add_argument( '--minTier1Mapq', help='path to tumor bam file', required = False )
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
78 parser.add_argument( '--minTier2Mapq', help='path to tumor bam file', required = False )
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
79 parser.add_argument( '--ssnvQuality_LowerBound', help='path to tumor bam file', required = False )
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
80 parser.add_argument( '--sindelQuality_LowerBound', help='path to tumor bam file', required = False )
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
81 parser.add_argument( '--isWriteRealignedBam', help='path to tumor bam file', required = False )
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
82 parser.add_argument( '--binSize', help='path to tumor bam file', required = False )
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
83 parser.add_argument( '--extraStrelkaArguments', help='path to tumor bam file', required = False )
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
84 parser.add_argument( '--isSkipDepthFilters', help='path to tumor bam file', required = False )
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
85 parser.add_argument( '--maxInputDepth', help='path to tumor bam file', required = False )
7
3e8541ece3c7 trying to fix environment variable 0.0.1
mini
parents: 6
diff changeset
86 parser.add_argument( '--scriptPath', help='path to tumor bam file', required = False )
6
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
87 args = parser.parse_args()
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
88
8
2cee4ed50058 nothing relevant
mini
parents: 7
diff changeset
89 root_dir= args.scriptPath
7
3e8541ece3c7 trying to fix environment variable 0.0.1
mini
parents: 6
diff changeset
90 expected_dir="for_tests"
3e8541ece3c7 trying to fix environment variable 0.0.1
mini
parents: 6
diff changeset
91 job_dir=os.getcwd()
3e8541ece3c7 trying to fix environment variable 0.0.1
mini
parents: 6
diff changeset
92 analysis_dir=job_dir + "/StrelkaAnalysis"
3e8541ece3c7 trying to fix environment variable 0.0.1
mini
parents: 6
diff changeset
93 config_script=root_dir + "/configureStrelkaWorkflow.pl"
9
0e8e6011082b Strelka Workflow 0.0.2
mini
parents: 8
diff changeset
94 tmp_dir = tempfile.mkdtemp( prefix='tmp-strelkaAnalysis-' )
7
3e8541ece3c7 trying to fix environment variable 0.0.1
mini
parents: 6
diff changeset
95 config_ini = "%s/config.ini" % (tmp_dir)
3e8541ece3c7 trying to fix environment variable 0.0.1
mini
parents: 6
diff changeset
96
14
ca84a74ff567 corriging bug in environment variable
mini
parents: 11
diff changeset
97 print("root_dir: " + root_dir + "\njob_dir :" + job_dir + "\nanalysis_dir :" + analysis_dir + "\nconfig_script :" + config_script + "\ntmp_dir :" + tmp_dir + "\nconfig_ini :" + config_ini)
7
3e8541ece3c7 trying to fix environment variable 0.0.1
mini
parents: 6
diff changeset
98
3e8541ece3c7 trying to fix environment variable 0.0.1
mini
parents: 6
diff changeset
99
6
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
100 #verifying eveything's ok
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
101 if not os.path.isfile(config_script):
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
102 sys.exit("ERROR: The strelka workflow must be built prior to running. See installation instructions in '$root_dir/README'")
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
103 print("configuring...", file=sys.stdout)
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
104 if os.path.exists(analysis_dir):
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
105 sys.exit("'" + analysis_dir + "' already exist, if you are executing this tool from galaxy it should not happen")
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
106
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
107
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
108 # creating index if needed
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
109 bam_filenames = [ args.tumorBam, args.normalBam ]
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
110 index_bam_files( bam_filenames, tmp_dir )
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
111 fasta_files = [ args.refFile ]
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
112 index_fasta_files( fasta_files, tmp_dir )
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
113
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
114 #creating config file if needed
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
115 if args.configFile == "Custom":
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
116 _create_config(vars(args), config_ini)
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
117 elif args.configFile == "Default":
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
118 cmdbash="cp %s %s" % (root_dir + "/strelka_config.sample", config_ini)
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
119 my_Popen(cmdbash, "copy_default_file_err", tmp_dir, "Error during the copy of default config file, maybe it was removed")
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
120 else:
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
121 if not os.path.exists(args.configFile):
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
122 print( "The path to your configuration File seems to be wrong, use another one or custom option", file=sys.stderr)
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
123 cmdbash="cp %s %s" % (args.configFile, config_ini)
9
0e8e6011082b Strelka Workflow 0.0.2
mini
parents: 8
diff changeset
124 my_Popen(cmdbash, "copy_default_file_err", tmp_dir, "Error during the copy of the selected config file")
6
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
125
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
126
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
127
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
128
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
129 #configuration of workflow
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
130 cmd="%s --tumor=%s --normal=%s --ref=%s --config=%s --output-dir=%s" % (config_script, args.tumorBam, args.normalBam, args.refFile, config_ini, analysis_dir)
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
131 print( "**** Starting configuration.")
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
132 print( "**** Configuration cmd: '" + cmd + "'")
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
133 my_Popen( cmd, "cinfugation_stderr", tmp_dir, "Error during configuration !")
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
134 print("completed configuration")
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
135
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
136 #run the workflow !
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
137 cmd="make -C " + analysis_dir
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
138 print("**** starting workflow.")
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
139 print("**** workflow cmd: '" + cmd + "'")
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
140 my_Popen( cmd, "workflow_stderr", tmp_dir, "Error during workflow execution !")
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
141 print("**** completed workflow execution")
87568e5a7d4f Testing strelka version 0.0.1
mini
parents:
diff changeset
142
14
ca84a74ff567 corriging bug in environment variable
mini
parents: 11
diff changeset
143 if __name__=='__main__':
ca84a74ff567 corriging bug in environment variable
mini
parents: 11
diff changeset
144 __main__()