annotate lefse2circlader.py @ 2:0efe0916b1b6 draft

Fix the kruskal wallis call error
author george-weingart
date Sat, 24 May 2014 13:27:35 -0400
parents 47ac77f2fe68
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
1 #!/usr/bin/env python
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
2
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
3 from __future__ import with_statement
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
4
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
5 import sys
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
6 import os
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
7 import argparse
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
8
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
9 def read_params(args):
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
10 parser = argparse.ArgumentParser(description='Convert LEfSe output to '
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
11 'Circlader input')
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
12 parser.add_argument( 'inp_f', metavar='INPUT_FILE', nargs='?',
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
13 default=None, type=str,
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
14 help="the input file [stdin if not present]")
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
15 parser.add_argument( 'out_f', metavar='OUTPUT_FILE', nargs='?',
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
16 default=None, type=str,
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
17 help="the output file [stdout if not present]")
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
18 parser.add_argument('-l', metavar='levels with label', default=0, type=int)
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
19
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
20 return vars(parser.parse_args())
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
21
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
22 def lefse2circlader(par):
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
23 finp,fout = bool(par['inp_f']), bool(par['out_f'])
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
24
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
25 with open(par['inp_f']) if finp else sys.stdin as inpf:
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
26 put_bm = (l.strip().split('\t') for l in inpf.readlines())
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
27 biomarkers = [p for p in put_bm if len(p) > 2]
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
28
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
29 circ = [ [ b[0],
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
30 "" if b[0].count('.') > par['l'] else b[0].split('.')[-1],
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
31 b[2],
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
32 b[2]+"_col" ] for b in biomarkers]
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
33
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
34 with open(par['out_f'],'w') if fout else sys.stdout as out_file:
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
35 for c in circ:
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
36 out_file.write( "\t".join( c ) + "\n" )
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
37
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
38 if __name__ == '__main__':
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
39 params = read_params(sys.argv)
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
40 lefse2circlader(params)
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
41
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
42
47ac77f2fe68 First version of lefse in this toolshed
george-weingart
parents:
diff changeset
43