view VCFStorage_wrapper.py @ 7:a6b557df86db draft

Uploaded
author urgi-team
date Tue, 15 Dec 2015 05:35:36 -0500
parents
children
line wrap: on
line source

#!/usr/bin/env python


import subprocess, tempfile, sys, os, glob, shutil, time, random
from optparse import OptionParser
from optparse import Option, OptionValueError

class VCFStorageWrapper(object):

	def __init__(self):
		self._options = None
		
		
	def stop_err(self, msg ):
		sys.stderr.write( "%s\n" % msg )
		sys.exit()
		
		
	def setAttributesFromCmdLine(self):
		description = "VCFStorage_wrapper"
		description += "\nWrapper for VCFStorage\n VCFStorage "
		description += "VCFStorage stores info from variant calling into a table. It will create a tabulate file with the different infos\n"
		description += "example : VCFStorage -f fasta.fa -l genomelist.list -w workdir -o output.tab \n"
		parser = OptionParser(description = description, version = "0.1")
		parser.add_option("-f", "--fasta",            dest = "fastaFile",      action = "store",     type = "string", help = "Input fasta file name [compulsory] [format: Fasta]",                default = "")
		parser.add_option("-l", "--genomeNameList",   dest = "genomeNameList", action = "append",    type = "string", help = "Input list of genome name ")
		parser.add_option("-L", "--genomeFileList",   dest = "genomeFileList", action = "append",    type = "string", help = "Input list of genome VCF file ")
		parser.add_option("-w", "--workDir",          dest = "workDir",        action = "store",     type = "string", help = "name of the workingDirectory",                                      default = "")
		parser.add_option("-o", "--out",              dest = "outFile",        action = "store",     type = "string", help = "Output file name [compulsory] [format: tab]",                       default = "")
		options = parser.parse_args()[0]
		self._setAttributesFromOptions(options)


	def _setAttributesFromOptions(self, options):
		self._options = options

	def run(self):
		self.createGenomeList()
		cmd = self.createCmdLine()
		self.launchCmdLine(cmd)
		
	def createGenomeList(self):
		self.genomelistFile = "%s.genomeListFile" % self._options.outFile
		lGenomeName = self._options.genomeNameList
		lGenomeVCF = self._options.genomeFileList
		output = open(self.genomelistFile, "w")
		for i,genomeName in enumerate(lGenomeName) :
			output.write("%s\t%s\n" % (lGenomeName[i],lGenomeVCF[i]))
		output.close()
		
	def createCmdLine(self):
		workdir = "VCFStorage_%s_%d" % (time.strftime("%d%H%M%S"), random.randint(0, 10000))
		prg = "VCFStorage.py"
		args = ""
		args += "-f %s" % self._options.fastaFile
		args += " "
		args += "-o %s" % self._options.outFile
		args += " "
		args += "-l %s" % self.genomelistFile
		args += " "
		args += "-w %s" % workdir
		cmd = "%s %s" %(prg, args)
		
		print cmd
		return cmd
		
	def launchCmdLine(self, cmd):
		try:
			tmp_err = tempfile.NamedTemporaryFile().name
			tmp_stderr = open( tmp_err, 'wb' )
			proc = subprocess.Popen( args=cmd, shell=True, cwd=".", stderr=tmp_stderr )
			returncode = proc.wait()
			tmp_stderr.close()
			# get stderr, allowing for case where it's very large
			tmp_stderr = open( tmp_err, 'rb' )
			stderr = ''
			buffsize = 1048576
			try:
				while True:
					stderr += tmp_stderr.read( buffsize )
					if not stderr or len( stderr ) % buffsize != 0:
						break
			except OverflowError:
				pass
			tmp_stderr.close()
			if stderr:
				raise Exception, stderr
		except Exception, e:
			os.remove(self.genomelistFile)
			self.stop_err( 'Error in VCFStorage:\n' + str( e ) ) 
		os.remove(self.genomelistFile)

if __name__ == "__main__":
	iWrapper = VCFStorageWrapper()
	iWrapper.setAttributesFromCmdLine()
	iWrapper.run()