dataLoading.py 文件源码

def parseDemultiplexConfig( demultiplexConfigFile ):
    """Parses the DemultiplexConfig file for the version of casava used

        Output is the details of the software in the format:
        [softwareName, softwareVersion, softwareArguements]
        [bcl2fastq, 1.8.4, "cluster/gsu/data/hiseq/140506_SN7001398_0099_BC4362ACXX/Data/Intensities/BaseCalls --output-dir /cluster/gsu/data/processed/hiseq/140506_SN7001398_0099_BC4362ACXX/ --sample-sheet /homes/gsupipe-x/SampleSheets/140506_SN7001398_0099_BC4362ACXX/SampleSheet_revComp_edited.csv --use-bases-mask Y*,I8n*,Y*'"]

    """
    casavaDetails = []
    try:
        print "Parsing Run Parameters"
        demultiplexConfigXML = ElementTree.parse(demultiplexConfigFile)
    except IOError:
        print "Cannot load information from %s" % demultiplexConfigFile
        raise
    except ElementTree.ParseError:
        print "Invalid XML in %s" % demultiplexConfigFile
        raise

    for softwareDetails in demultiplexConfigXML.iterfind("Software"):
        versionString = softwareDetails.attrib["Version"]
        commandArgs = softwareDetails.attrib["CmdAndArgs"]

    casavaDetails.append( versionString.split("-")[0] )
    casavaDetails.append( versionString.split("-")[1] )
    casavaDetails.append( commandArgs )

    return casavaDetails


   # for line in demultiplexConfigFile: