Mercurial > repos > urgi-team > gandalfworkflow

comparison VCFStorage_wrapper.py @ 7:a6b557df86db draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded
author urgi-team
date Tue, 15 Dec 2015 05:35:36 -0500
parents
children
comparison
equal deleted inserted replaced
6:7557b71a92e7 7:a6b557df86db
1 #!/usr/bin/env python
2
3
4 import subprocess, tempfile, sys, os, glob, shutil, time, random
5 from optparse import OptionParser
6 from optparse import Option, OptionValueError
7
8 class VCFStorageWrapper(object):
9
10 def __init__(self):
11 self._options = None
12
13
14 def stop_err(self, msg ):
15 sys.stderr.write( "%s\n" % msg )
16 sys.exit()
17
18
19 def setAttributesFromCmdLine(self):
20 description = "VCFStorage_wrapper"
21 description += "\nWrapper for VCFStorage\n VCFStorage "
22 description += "VCFStorage stores info from variant calling into a table. It will create a tabulate file with the different infos\n"
23 description += "example : VCFStorage -f fasta.fa -l genomelist.list -w workdir -o output.tab \n"
24 parser = OptionParser(description = description, version = "0.1")
25 parser.add_option("-f", "--fasta", dest = "fastaFile", action = "store", type = "string", help = "Input fasta file name [compulsory] [format: Fasta]", default = "")
26 parser.add_option("-l", "--genomeNameList", dest = "genomeNameList", action = "append", type = "string", help = "Input list of genome name ")
27 parser.add_option("-L", "--genomeFileList", dest = "genomeFileList", action = "append", type = "string", help = "Input list of genome VCF file ")
28 parser.add_option("-w", "--workDir", dest = "workDir", action = "store", type = "string", help = "name of the workingDirectory", default = "")
29 parser.add_option("-o", "--out", dest = "outFile", action = "store", type = "string", help = "Output file name [compulsory] [format: tab]", default = "")
30 options = parser.parse_args()[0]
31 self._setAttributesFromOptions(options)
32
33
34 def _setAttributesFromOptions(self, options):
35 self._options = options
36
37 def run(self):
38 self.createGenomeList()
39 cmd = self.createCmdLine()
40 self.launchCmdLine(cmd)
41
42 def createGenomeList(self):
43 self.genomelistFile = "%s.genomeListFile" % self._options.outFile
44 lGenomeName = self._options.genomeNameList
45 lGenomeVCF = self._options.genomeFileList
46 output = open(self.genomelistFile, "w")
47 for i,genomeName in enumerate(lGenomeName) :
48 output.write("%s\t%s\n" % (lGenomeName[i],lGenomeVCF[i]))
49 output.close()
50
51 def createCmdLine(self):
52 workdir = "VCFStorage_%s_%d" % (time.strftime("%d%H%M%S"), random.randint(0, 10000))
53 prg = "VCFStorage.py"
54 args = ""
55 args += "-f %s" % self._options.fastaFile
56 args += " "
57 args += "-o %s" % self._options.outFile
58 args += " "
59 args += "-l %s" % self.genomelistFile
60 args += " "
61 args += "-w %s" % workdir
62 cmd = "%s %s" %(prg, args)
63
64 print cmd
65 return cmd
66
67 def launchCmdLine(self, cmd):
68 try:
69 tmp_err = tempfile.NamedTemporaryFile().name
70 tmp_stderr = open( tmp_err, 'wb' )
71 proc = subprocess.Popen( args=cmd, shell=True, cwd=".", stderr=tmp_stderr )
72 returncode = proc.wait()
73 tmp_stderr.close()
74 # get stderr, allowing for case where it's very large
75 tmp_stderr = open( tmp_err, 'rb' )
76 stderr = ''
77 buffsize = 1048576
78 try:
79 while True:
80 stderr += tmp_stderr.read( buffsize )
81 if not stderr or len( stderr ) % buffsize != 0:
82 break
83 except OverflowError:
84 pass
85 tmp_stderr.close()
86 if stderr:
87 raise Exception, stderr
88 except Exception, e:
89 os.remove(self.genomelistFile)
90 self.stop_err( 'Error in VCFStorage:\n' + str( e ) )
91 os.remove(self.genomelistFile)
92
93 if __name__ == "__main__":
94 iWrapper = VCFStorageWrapper()
95 iWrapper.setAttributesFromCmdLine()
96 iWrapper.run()