Example #1
0
def main():
    description = """
    Take a blast result table and output a subset of hits based on the
    chosen filtering options. If more than one blast file given, use -O
    to get multiple output files, otherwise all output data will be
    concatenated into one output.
    """

# command line arguments
    parser = argparse.ArgumentParser(
        description=description,
        conflict_handler='resolve')
    add_hit_table_arguments(parser, flags='all')
    parser.add_argument(
        "-o",
        "--outfilenome",
        dest="outfilename",
        default=None,
        metavar="OUTFILENAME",
        help="Write masked fasta output to OUTFILENAME.")
    parser.add_argument(
        '-O',
        '--autoOutName',
        default=False,
        action='store_true',
        help="Automatically generate output file name from input name "
             "and options. Overridden by -o, cannot be used with data "
             "from STDIN.")
    parser.add_argument('-G', '--gff', default=False, action='store_true',
                        help="output GFF format instead of input format")
    parser.add_argument('hit_table', nargs='*',
                        type=argparse.FileType('rU'), default=[sys.stdin, ],
                        help="Table of search results to be filtered. "
                             "If absent, data will be read from STDIN")

    add_universal_arguments(parser)

    arguments = parser.parse_args()

    setup_logging(arguments)

    # check that we have blast file as argument

    # if we're not doing auto file names, wriate all outputs to same file
    if not arguments.autoOutName:
        if arguments.outfilename is not None:
            logging.info("Writing data to %s" % (arguments.outfilename))
            outfile_handle = open(arguments.outfilename, 'w')
        else:
            logging.info("writing data to STDOUT")
            outfile_handle = sys.stdout

    if arguments.gff:
        logging.info("Converting to GFF")

    # loop over inputs
    for infile_handle in arguments.hit_table:
        logging.info("reading data from %s" % (infile_handle.name))
        if arguments.autoOutName:
            outfile_handle = open(
                getOutputFile(
                    infile_handle.name,
                    arguments),
                'w')

        # filter
        params = FilterParams.create_from_arguments(arguments)
        filterM8(infile_handle, outfile_handle, params, to_gff=arguments.gff)

        if arguments.autoOutName:
            outfile_handle.close()
        infile_handle.close()
Example #2
0
def main():
    description = """
    Take a blast result table and output a subset of hits based on the
    chosen filtering options. If more than one blast file given, use -O
    to get multiple output files, otherwise all output data will be
    concatenated into one output.
    """

    # command line arguments
    parser = argparse.ArgumentParser(description=description,
                                     conflict_handler='resolve')
    add_hit_table_arguments(parser, flags='all')
    parser.add_argument("-o",
                        "--outfilenome",
                        dest="outfilename",
                        default=None,
                        metavar="OUTFILENAME",
                        help="Write masked fasta output to OUTFILENAME.")
    parser.add_argument(
        '-O',
        '--autoOutName',
        default=False,
        action='store_true',
        help="Automatically generate output file name from input name "
        "and options. Overridden by -o, cannot be used with data "
        "from STDIN.")
    parser.add_argument('-G',
                        '--gff',
                        default=False,
                        action='store_true',
                        help="output GFF format instead of input format")
    parser.add_argument('hit_table',
                        nargs='*',
                        type=argparse.FileType('rU'),
                        default=[
                            sys.stdin,
                        ],
                        help="Table of search results to be filtered. "
                        "If absent, data will be read from STDIN")

    add_universal_arguments(parser)

    arguments = parser.parse_args()

    setup_logging(arguments)

    # check that we have blast file as argument

    # if we're not doing auto file names, wriate all outputs to same file
    if not arguments.autoOutName:
        if arguments.outfilename is not None:
            logging.info("Writing data to %s" % (arguments.outfilename))
            outfile_handle = open(arguments.outfilename, 'w')
        else:
            logging.info("writing data to STDOUT")
            outfile_handle = sys.stdout

    if arguments.gff:
        logging.info("Converting to GFF")

    # loop over inputs
    for infile_handle in arguments.hit_table:
        logging.info("reading data from %s" % (infile_handle.name))
        if arguments.autoOutName:
            outfile_handle = open(getOutputFile(infile_handle.name, arguments),
                                  'w')

        # filter
        params = FilterParams.create_from_arguments(arguments)
        filterM8(infile_handle, outfile_handle, params, to_gff=arguments.gff)

        if arguments.autoOutName:
            outfile_handle.close()
        infile_handle.close()
def main():
    usage = "usage: %prog [OPTIONS] BLAST_FILE"
    description = """
    Take a blast result table and output a subset of hits based on the chosen filtering options. If more than one blast file given, use -O to get multiple output files, otherwise all output data will be concatenated into one output.
    """

# command line options
    parser = OptionParser(usage, description=description, conflict_handler='resolve')
    addHitTableOptions(parser, flags='all')
    parser.add_option("-o", "--outfilenome", dest="outfilename", default=None,
                      metavar="OUTFILENAME", help="Write masked fasta output to OUTFILENAME.")
    parser.add_option('-O', '--autoOutName', default=False,
                      action='store_true',
                      help="Automatically generate output file name from input name and options. Overridden by -o, cannot be used with data from STDIN.")

    addUniversalOptions(parser)

    (options, args) = parser.parse_args()

    setupLogging(options,description)

    if options.hitTableFormat=='last':
        if options.hitTableSort=='evalue':
            parser.error("The last format has no evalue to sort by, sorry")

    # check that we have blast file as argument
    if len(args) <= 1:
        # input
        if len(args) == 1:
            infile = args[0]
            logging.info("reading data from %s" % (infile))
            instream = open(infile,'rU')
        else:
            infile = './stdin'
            logging.info("reading data from STDIN")
            instream=sys.stdin

        # output
        if options.outfilename is not None:
            logging.info("Writing data to %s" % (options.outfilename))
            outstream=open(options.outfilename,'w')
        elif options.autoOutName:
            outfile=getOutputFile(infile,options)
            logging.info("Writing data to %s" % (outfile))
            outstream=open(outfile,'w')
        else:
            logging.info("writing data to STDOUT")
            outstream=sys.stdout

        # filter
        params=FilterParams.createFromOptions(options)
        filterM8(instream,outstream,params)
    else:
        if not options.autoOutName:
            if options.outfilename is not None:
                logging.info("Writing data to %s" % (options.outfilename))
                outstream=open(options.outfilename,'w')
            else:
                logging.info("writing data to STDOUT")
                outstream=sys.stdout
        for infilename in args:
            logging.info("reading data from %s" % (infilename))
            instream=open(infilename,'rU')
            if options.autoOutName:
                outstream=open(getOutputFile(infilename,options),'w')

            # filter
            params=FilterParams.createFromOptions(options)
            filterM8(instream,outstream,params)

            if options.autoOutName:
                outstream.close()
            instream.close()