Transfrom content of Kallisto matrix, used for in silico ribotyping of C. difficile in this case

transform_kallisto_bacseq.py

#!/usr/bin/env python

"""Transform ribotype data. Input matrix
is a transposed output from bac_seq"""

from __future__ import print_function
from __future__ import division
import sys
import os
import optparse

def test_file(option, opt_str, value, parser):
    try:
        with open(value): setattr(parser.values, option.dest, value)
    except IOError:
        print('%s file cannot be opened' % option)
        sys.exit()

def transpose_matrix(matrix):
    out_matrix = open("transposed_matrix.matrix", "w")
    reduced = [ ]
    for line in open(matrix, "rU"):
        newline=line.strip("\n")
        fields = newline.split("\t")
        reduced.append(fields)
    test=map(list, zip(*reduced))
    for x in test:
        out_matrix.write("\t".join(x)+"\n")
    out_matrix.close()

def main(matrix, min_percent):
    transpose_matrix(matrix)
    print("Genome hits:")
    for line in open("transposed_matrix.matrix", "rU"):
        newline = line.strip()
        fields = newline.split()
        if fields[0] == "target":
            pass
        else:
            values = []
            to_write = []
            for x in fields[1:]:
                values.append(int(x))
            max_value = max(values)
            for value in values:
                try:
                    if float(value)/max_value > float(min_percent):
                        to_write.append("1")
                    else:
                        to_write.append("0")
                except:
                    to_write.append("0")
            if "0" in to_write[1:]:
                pass
            else:
                print(fields[0])
    os.system("rm transposed_matrix.matrix")

if __name__ == "__main__":
    usage="usage: %prog [options]"
    parser = optparse.OptionParser(usage=usage)
    parser.add_option("-m", "--matrix", dest="matrix",
                      help="/path/to/kallisto_merged_matrix [REQUIRED]",
                      type="string", action="callback", callback=test_file)
    parser.add_option("-p", "--min_percent", dest="min_percent",
                      help="minimum percent of maximum to be called present, defaults to 0.2",
                      type="float", action="store", default="0.2")
    options, args = parser.parse_args()

    mandatories = ["matrix"]
    for m in mandatories:
        if not getattr(options, m, None):
            print("\nMust provide %s.\n" %m)
            parser.print_help()
            exit(-1)
    main(options.matrix,options.min_percent)