Usage: python extract_bcs_from_fastq.py X Y Z A B where: X is input fastq file Y is output barcode reads fastq file Z is output reads (with barcode removed) fastq file A is size of barcode B is True/False for reverse complement of barcode before writing

extract_bcs_from_fastq.py

#!/usr/bin/env python

from sys import argv

from cogent.parse.fastq import MinimalFastqParser
from cogent import DNA



f = open(argv[1], "U")

bc_out = open(argv[2], "w")
read_out = open(argv[3], "w")

barcode_size = int(argv[4])

if argv[5] == "True":
    rc_bc = True
elif argv[5] == "False":
    rc_bc = False
else:
    raise ValueError,("Fifth argument must be 'True' or 'False' to set"
     " reverse complement of barcodes.")

for data in MinimalFastqParser(f, strict=False):
   curr_label = data[0].strip()
   curr_bc_read = data[1].strip()[0:barcode_size]
   curr_r3_read = data[1].strip()[barcode_size:]
   curr_bc_qual = data[2].strip()[0:barcode_size]
   curr_r3_qual = data[2].strip()[barcode_size:]
   bc_out.write("@%s\n" % curr_label)
   if rc_bc:
      curr_bc_read = DNA.rc(curr_bc_read)
   bc_out.write("%s\n" % curr_bc_read)
   bc_out.write("+\n")
   bc_out.write("%s\n" % curr_bc_qual)
   read_out.write("@%s\n" % curr_label)
   read_out.write("%s\n" % curr_r3_read)
   read_out.write("+\n")
   read_out.write("%s\n" % curr_r3_qual)