view VCFGandalfTools/VCFCarto_wrapper.py @ 3:1fd1f727c330 draft default tip

Uploaded

#!/usr/bin/env python


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


class VCFCartoWrapper(object):

    def __init__(self):
        self._options = None
        
        
    def stop_err(self, msg ):
        sys.stderr.write( "%s\n" % msg )
        sys.exit()
        
        
    def setAttributesFromCmdLine(self):
        description = "VCFcarto can convert your tabulated file into a file with only the SNP from refA and refH.\n"
        description += "2 formats are possible, either the input format is conserved, or the format is changed into a 3 letter format\n"
        description += "(\"A\" for refA, \"H\" for refH and \"-\" when the base do not correspond to any parent)\n\n"
        description += "example 1 : VCFcarto.py -f Storage.out -A G15 -H G23 -o FilteredStorage.out\n"
        description += "example 2 : VCFcarto.py -f Storage.out -A ref1 -H ref2 -p -s -g -m -o cartoTable.out\n"
        parser = OptionParser(description = description, version = "0.2") 
        parser.add_option("-f", "--file",      dest = "tableName", action = "store", type = "string", help = "Input TSV File name [compulsory] [format: TSV]",                                                                                                    default = "")
        parser.add_option("-o", "--output",    dest = "outFile",   action = "store", type = "string", help = "output TSV File name [compulsory] [format: TSV]",                                                                                                   default = "")
        parser.add_option("-A", "--refA",      dest = "refA",      action = "store", type = "string", help = "name of the reference genome A [compulsory] ",                                                                                                      default = "")
        parser.add_option("-H", "--refH",      dest = "refH",      action = "store", type = "string", help = "name of the reference genome H [compulsory] ",                                                                                                      default = "")
        parser.add_option("-p", "--onlyParents",   dest = "onlyPar",   action = "store_true",         help = "Will change every letters by either A or H depending on which parents the strain correspond to for that base[optional] [default: False]",           default = False)
        parser.add_option("-m", "--mergeMarkers",   dest = "mergeMarkers",   action = "store_true",   help = "Will merge sequential markers with the same information ; option -p is needed [optional] [default: False]",                                         default = False)
        parser.add_option("-M", "--mergeFile", dest = "mergeFile", action = "store", type = "string", help = "name for the mergeFile ",                                                                                                                           default = "")
        parser.add_option("-g", "--graphics",  dest = "graphs",    action = "store_true",             help = "create graphs. Only works with -p[optional] [default: False]",                                                                                      default = False)
        parser.add_option("-G", "--graphHTML", dest = "graphHTML", action = "store", type = "string", help = "name of the HTML linking to graphs ",                                                                                                              default = "")
        parser.add_option("-d", "--dirGraphs", dest = "dirGraphs", action = "store", type = "string", help = "name of the folder containing graphs ",                                                                                                              default = "")
        options = parser.parse_args()[0]
        self._setAttributesFromOptions(options)


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

    def run(self):
        prg = "VCFCarto.py"
        args = ""
        args += "-f %s" % self._options.tableName
        args += " "
        args += "-o %s" % self._options.outFile
        args += " "
        args += "-v 2 "
        if self._options.onlyPar :
            args += " "
            args += "-p"
            if self._options.mergeMarkers :
                args += " "
                args += "-m"
            if self._options.graphs :
                args += " "
                args += "-g"
        else : 
            args += " "
            args += "-A %s" % self._options.refA
            args += " "
            args += "-H %s" % self._options.refH
        cmd = "%s %s" %(prg, args)
        
        print 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:
            self.stop_err( 'Error in VCFCarto:\n' + str( e ) ) 
        
        if self._options.mergeMarkers :
            shutil.move("markerList.bed" ,self._options.mergeFile)
        if self._options.graphs :
            html = open(self._options.graphHTML, "w")
            
            os.mkdir(self._options.dirGraphs)
            lGraphsFiles = glob.glob("VCFCarto_graphs/*")
            for file in lGraphsFiles :
                baseName = os.path.basename(file)
                shutil.move( file ,"%s/%s" %(self._options.dirGraphs, baseName))
                line = "<img src=\"%s\" > \n" %(baseName)
                html.write(line)
                
if __name__ == "__main__":
    iWrapper = VCFCartoWrapper()
    iWrapper.setAttributesFromCmdLine()
    iWrapper.run()