1 | #!/usr/bin/env python |
---|
2 | |
---|
3 | """ |
---|
4 | Split fixed length paired end reads |
---|
5 | """ |
---|
6 | |
---|
7 | import os, sys |
---|
8 | |
---|
9 | if __name__ == '__main__': |
---|
10 | |
---|
11 | infile = sys.argv[1] |
---|
12 | outfile_end1 = open(sys.argv[2], 'w') |
---|
13 | outfile_end2 = open(sys.argv[3], 'w') |
---|
14 | |
---|
15 | i = 0 |
---|
16 | |
---|
17 | for line in file( infile ): |
---|
18 | line = line.rstrip() |
---|
19 | |
---|
20 | if not line: |
---|
21 | continue |
---|
22 | |
---|
23 | end1 = '' |
---|
24 | end2 = '' |
---|
25 | |
---|
26 | line_index = i % 4 |
---|
27 | |
---|
28 | if line_index == 0: |
---|
29 | end1 = line + '/1' |
---|
30 | end2 = line + '/2' |
---|
31 | |
---|
32 | elif line_index == 1: |
---|
33 | seq_len = len(line)/2 |
---|
34 | end1 = line[0:seq_len] |
---|
35 | end2 = line[seq_len:] |
---|
36 | |
---|
37 | elif line_index == 2: |
---|
38 | end1 = line + '/1' |
---|
39 | end2 = line + '/2' |
---|
40 | |
---|
41 | else: |
---|
42 | qual_len = len(line)/2 |
---|
43 | end1 = line[0:qual_len] |
---|
44 | end2 = line[qual_len:] |
---|
45 | |
---|
46 | outfile_end1.write('%s\n' %(end1)) |
---|
47 | outfile_end2.write('%s\n' %(end2)) |
---|
48 | |
---|
49 | i += 1 |
---|
50 | |
---|
51 | if i % 4 != 0 : |
---|
52 | sys.stderr.write("WARNING: Number of lines in the input file was not divisible by 4.\nCheck consistency of the input fastq file.\n") |
---|
53 | outfile_end1.close() |
---|
54 | outfile_end2.close() |
---|