aggregate_matrix.py

#!/bin/env python

import sys
import getopt


def matrix_iterator(instream=sys.stdin):
    """Iterate over the rows of the matrix."""
    for line in instream:

        # remove the endline
        line = line.rstrip()

        # skip empty line
        if len(line) == 0:
            continue

        # return the fields
        yield line.split("\t")


def main(instream=sys.stdin, outstream=sys.stdout):
    """The main program loop."""
    base = None
    counts = []

    # foreach entry in the array
    for columns in matrix_iterator(instream):

        # first entry
        if base is None:
            base = columns[:7]

        # if we are changing fragment
        if base[:7] != columns[:7]:
            outstream.write("%s\t%d\n" % (
                "\t".join(base), sum(counts)))
            base = columns[:7]
            counts = []

        # add the score
        try:
            counts.append(int(columns[7]))
        except ValueError:
            counts.append(0)

    # write the output entry
    outstream.write("%s\t%d\n" % ("\t".join(base), sum(counts)))


if __name__ == "__main__":

    def usage(message="", error=None):
        """Print the usage information."""
        sys.stdout.write("""
Message %(message)s

Called as %(commandline)s

Usage:
    python %(tool)s -i [input file] -o [output file] 

Options:
    -i/--input    [file]  the matrix file, default stdin
    -o/--output   [file]  the output matrix file, default stdout
    -h/--help     []  print this message

""" % {
    "message": message,
    "tool": sys.argv[0],
    "commandline": " ".join(sys.argv)
})
        if error is not None:
            sys.exit(error)
        # end of usage

    # main variables
    fstream_input = sys.stdin
    fstream_output = sys.stdout

    # option parsing
    shortopt = "i:o:h"
    longopt = ["input=", "output=", "help"]

    try:
        opts, args = getopt.getopt(sys.argv[1:], shortopt, longopt)
        for o, a in opts:
            if o in ("-i", "--input"):
                fstream_input = open(a, "r")
            elif o in ("-o", "--output"):
                fstream_output = open(a, "w")
            elif o in ('-h', '--help'):
                usage("Help was asked", error=0)
    except getopt.GetoptError as err:
        usage(str(err), error=2)

    # run the program loop
    main(fstream_input, fstream_output)

    if not fstream_input.closed and fstream_input is not sys.stdin:
        fstream_input.close()

    if not fstream_output.closed and fstream_output is not sys.stdout:
        fstream_output.close()