get_gipz_counts.py

import argparse
import sys
from Bio import SeqIO

def do_count(passengers, alignments, countfile):
    counts = initialize_counts(passengers)
    for line in alignments:
        tokens = line.rstrip().split('\t')
        if tokens[2] != '*':
            counts[tokens[2]] += 1
    for key in counts.keys():
        countfile.write(key+'\t'+str(counts[key])+'\n')

def initialize_counts(pool):
    counts = {}
    records = SeqIO.parse(pool, 'fasta')
    for record in records:
        counts[record.id] = 0
    return counts

if __name__ == '__main__':
    parser = argparse.ArgumentParser(description='Assign reads to GIPZ library constructs')
    parser.add_argument('passengers', type=argparse.FileType('r'), help='FASTA sequence file of GIPZ Passengers (generated by get_gipz_passengers.py')
    parser.add_argument('infile', nargs='?', type=argparse.FileType('r'), default=sys.stdin, help="Optional input file of alignments (or read from STDIN)")
    parser.add_argument('outfile', nargs='?', type=argparse.FileType('w'), default=sys.stdout, help="Optional output file of read counts (or write to STDOUT)")
    args = parser.parse_args()
    do_count(args.passengers, args.infile, args.outfile)