Usage: python split_fasta_qual_seqs.py X Y Z A B C D where X - input fasta file Y - input qual file Z - Number of seqs to write to first file, remaining will be written to second A - first output fasta file B - second output fasta file C - first output qual file D - second output qual file One can use a: grep -c "^>" fasta_filepath to see how ma…

split_fasta_qual_seqs.py

#!/usr/bin/env python

""" Usage:
python split_fasta_qual_seqs.py X Y Z A B C D
where
X - input fasta file
Y - input qual file
Z - Number of seqs to write to first file, remaining will be written to second
A - first output fasta file
B - second output fasta file
C - first output qual file
D - second output qual file
"""

from sys import argv

in_fasta = open(argv[1], "U")
in_qual = open(argv[2], "U")
seq_count = int(argv[3])
out_fasta1 = open(argv[4], "w")
out_fasta2 = open(argv[5], "w")
out_qual1 = open(argv[6], "w")
out_qual2 = open(argv[7], "w")

counter = 0
for line in in_fasta:
    if line.startswith(">"):
        counter += 1
    if counter > seq_count:
        out_fasta2.write(line)
    else:
        out_fasta1.write(line)
    
    
counter = 0
for line in in_qual:
    if line.startswith(">"):
        counter += 1
    if counter > seq_count:
        out_qual2.write(line)
    else:
        out_qual1.write(line)


in_fasta.close()
in_qual.close()
out_fasta1.close()
out_fasta2.close()
out_qual1.close()
out_qual2.close()