annotate prince_postprocess.py @ 0:88748989a22a draft default tip

planemo upload
author jowong
date Tue, 13 Nov 2018 11:41:37 -0500
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
88748989a22a planemo upload
jowong
parents:
diff changeset
1 #!/usr/bin/env python
88748989a22a planemo upload
jowong
parents:
diff changeset
2
88748989a22a planemo upload
jowong
parents:
diff changeset
3 import sys
88748989a22a planemo upload
jowong
parents:
diff changeset
4 import argparse as ap
88748989a22a planemo upload
jowong
parents:
diff changeset
5 import re
88748989a22a planemo upload
jowong
parents:
diff changeset
6 parser = ap.ArgumentParser(prog='prince_postprocess', conflict_handler='resolve',
88748989a22a planemo upload
jowong
parents:
diff changeset
7 description="Postprocess galaxy PRINCE output")
88748989a22a planemo upload
jowong
parents:
diff changeset
8
88748989a22a planemo upload
jowong
parents:
diff changeset
9 input = parser.add_argument_group('Input', '')
88748989a22a planemo upload
jowong
parents:
diff changeset
10 input.add_argument('-i', '--input', nargs=1, required=True, help="PRINCE OUTPUT")
88748989a22a planemo upload
jowong
parents:
diff changeset
11 input.add_argument('-s', '--sample', nargs='+', required=True, help="Sample names")
88748989a22a planemo upload
jowong
parents:
diff changeset
12 input.add_argument('-f', '--file', nargs='+', required=True, help="File of forward reads in galaxy convention")
88748989a22a planemo upload
jowong
parents:
diff changeset
13 if len(sys.argv) == 0:
88748989a22a planemo upload
jowong
parents:
diff changeset
14 parser.print_usage()
88748989a22a planemo upload
jowong
parents:
diff changeset
15 sys.exit(1)
88748989a22a planemo upload
jowong
parents:
diff changeset
16
88748989a22a planemo upload
jowong
parents:
diff changeset
17 args = parser.parse_args()
88748989a22a planemo upload
jowong
parents:
diff changeset
18
88748989a22a planemo upload
jowong
parents:
diff changeset
19 #print(args.input)
88748989a22a planemo upload
jowong
parents:
diff changeset
20 #sample_name = re.sub('(_1.fastq(.gz)*|_2.fastq(.gz)*|.fastq(.gz)*)', '', args.label.rstrip().lstrip())
88748989a22a planemo upload
jowong
parents:
diff changeset
21 sample_dict={}
88748989a22a planemo upload
jowong
parents:
diff changeset
22 path_index = 0
88748989a22a planemo upload
jowong
parents:
diff changeset
23 for path in args.file:
88748989a22a planemo upload
jowong
parents:
diff changeset
24 with open(path) as f:
88748989a22a planemo upload
jowong
parents:
diff changeset
25 for line in f:
88748989a22a planemo upload
jowong
parents:
diff changeset
26 sample_dict[re.sub(".*/","",line)] = sample[path_index]
88748989a22a planemo upload
jowong
parents:
diff changeset
27 path_index += 1
88748989a22a planemo upload
jowong
parents:
diff changeset
28
88748989a22a planemo upload
jowong
parents:
diff changeset
29
88748989a22a planemo upload
jowong
parents:
diff changeset
30
88748989a22a planemo upload
jowong
parents:
diff changeset
31
88748989a22a planemo upload
jowong
parents:
diff changeset
32
88748989a22a planemo upload
jowong
parents:
diff changeset
33
88748989a22a planemo upload
jowong
parents:
diff changeset
34
88748989a22a planemo upload
jowong
parents:
diff changeset
35 with open(args.input[0]) as prince_output:
88748989a22a planemo upload
jowong
parents:
diff changeset
36 with open('prince_postprocess_output.txt', 'w') as output:
88748989a22a planemo upload
jowong
parents:
diff changeset
37 x = 1
88748989a22a planemo upload
jowong
parents:
diff changeset
38 index = 0
88748989a22a planemo upload
jowong
parents:
diff changeset
39 for line in prince_output:
88748989a22a planemo upload
jowong
parents:
diff changeset
40 if x%2 == 0:
88748989a22a planemo upload
jowong
parents:
diff changeset
41 entries =line.rstrip().split(',')
88748989a22a planemo upload
jowong
parents:
diff changeset
42 sample = re.sub('(_1.fastq(.gz)*|_2.fastq(.gz)*|.fastq(.gz)*)', '', sample_dict[entries[0]])
88748989a22a planemo upload
jowong
parents:
diff changeset
43 output.write(re.sub(entries[0], sample, line))
88748989a22a planemo upload
jowong
parents:
diff changeset
44 index += 1
88748989a22a planemo upload
jowong
parents:
diff changeset
45 else:
88748989a22a planemo upload
jowong
parents:
diff changeset
46 output.write(line)
88748989a22a planemo upload
jowong
parents:
diff changeset
47 x += 1
88748989a22a planemo upload
jowong
parents:
diff changeset
48 #output.write("\n")
88748989a22a planemo upload
jowong
parents:
diff changeset
49
88748989a22a planemo upload
jowong
parents:
diff changeset
50