annotate parse_strelka_makefile.py @ 0:88141dfd5db1 draft

Uploaded
author morinlab
date Thu, 18 Aug 2016 19:34:55 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
1 import argparse
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
2 import sys
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
3 import re
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
4
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
5 if __name__ == '__main__':
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
6
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
7 desc = "Convert Strelka Makefile into a sub-Makefile used STRICTLY for parallelization by chromosome in Galaxy"
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
8 parser = argparse.ArgumentParser(description=desc)
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
9 parser.add_argument(
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
10 "-m", "--makefile",
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
11 required=True,
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
12 help="Input Must be a Strelka Makefile"
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
13 )
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
14 parser.add_argument(
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
15 "-c", "--chrom",
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
16 type=str,
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
17 required=False,
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
18 help="Chromosome present in the Strelka Makefile"
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
19 )
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
20 parser.add_argument(
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
21 "-cf", "--chrom_file",
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
22 required=False,
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
23 help="Chromosome File separated by newline"
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
24 )
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
25 parser.add_argument(
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
26 '-o', '--output',
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
27 required=True,
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
28 help="Output Makefile"
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
29 )
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
30 args = parser.parse_args()
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
31
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
32 if (not args.chrom and not args.chrom_file) or (args.chrom and args.chrom_file):
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
33 sys.exit('Must define a single chromosome entry or a single chromosome file')
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
34
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
35 chromosomes = [];
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
36
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
37 if args.chrom_file:
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
38 chromosomes = open(args.chrom_file).read().splitlines()
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
39
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
40 if args.chrom:
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
41 chromosomes = [args.chrom]
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
42 add_at_end = []
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
43 makefile_in = open(args.makefile, 'r')
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
44 makefile_out = open(args.output, 'w')
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
45
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
46 for line in makefile_in:
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
47 if line.startswith('script_dir'):
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
48 makefile_out.write(line)
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
49 elif line.startswith('call_script'):
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
50 makefile_out.write(line)
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
51 elif line.startswith('filter_script'):
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
52 makefile_out.write(line)
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
53 elif line.startswith('finish_script'):
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
54 makefile_out.write(line)
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
55 elif line.startswith('script_dir'):
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
56 makefile_out.write(line)
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
57 elif line.startswith('config_file'):
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
58 makefile_out.write(line)
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
59 elif line.startswith('analysis_dir'):
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
60 makefile_out.write(line)
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
61 elif line.startswith('results_dir'):
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
62 makefile_out.write(line)
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
63 elif line.startswith('get_chrom_dir'):
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
64 makefile_out.write(line)
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
65 elif line.startswith('get_chrom_task'):
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
66 makefile_out.write(line)
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
67 elif line.startswith('get_bin_task'):
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
68 makefile_out.write(line)
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
69 makefile_out.write('\n')
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
70 elif line.startswith('all:'):
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
71 makefile_out.write('all:\n')
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
72 elif re.search("--chrom=", line):
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
73 if re.search(''.join(["\s|".join(chromosomes),'\s']), line):
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
74 if re.search('bin', line):
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
75 makefile_out.write(line)
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
76 else:
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
77 add_at_end.append(line)
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
78
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
79 makefile_in.close()
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
80 for line in add_at_end:
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
81 makefile_out.write(line)
88141dfd5db1 Uploaded
morinlab
parents:
diff changeset
82 makefile_out.close()