Mercurial > repos > tomnl > track_rt_raw

annotate msp_split.py @ 26:95a4c22e0d64 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit a598583947118bd50cfe5ae12d955c20d6e502a6
author tomnl
date Tue, 17 Jul 2018 12:35:15 -0400
parents bd298a4f21e4
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
25
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
1 from __future__ import print_function
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
2 import argparse
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
3 import textwrap
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
4 import os
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
5 import re
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
6 import csv
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
7 import math
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
8
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
9 def msp_split(i, o, n):
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
10 spec_total = lcount('NAME', i)
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
11 spec_lim = math.ceil(spec_total/float(n))
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
12 spec_c = 0
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
13 filelist = []
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
14 header = ''
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
15 print('spec_lim', spec_lim)
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
16 with open(i, 'r') as msp_in:
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
17 for i in range(1, n+1):
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
18 with open(os.path.join(o, 'file{}.msp'.format(str(i).zfill(len(str(n))))), 'w+') as msp_out:
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
19 while spec_c <= spec_lim:
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
20 if header:
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
21 msp_out.write(header)
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
22 header = ''
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
23 line = msp_in.readline()
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
24
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
25 if not line:
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
26 break # end of file
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
27
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
28 if re.match('^NAME:.*$', line, re.IGNORECASE):
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
29 header = line
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
30 spec_c += 1
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
31 else:
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
32 msp_out.write(line)
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
33 spec_c = 1
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
34
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
35 return filelist
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
36
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
37 def lcount(keyword, fname):
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
38 with open(fname, 'r') as fin:
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
39 return sum([1 for line in fin if keyword in line])
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
40
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
41 def main():
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
42
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
43 p = argparse.ArgumentParser(prog='PROG',
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
44 formatter_class=argparse.RawDescriptionHelpFormatter,
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
45 description='''split msp files''',
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
46 )
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
47
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
48 p.add_argument('-i', dest='i', help='msp file', required=True)
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
49 p.add_argument('-o', dest='o', help='out dir', required=True)
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
50 p.add_argument('-n', dest='n',)
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
51
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
52
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
53 args = p.parse_args()
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
54
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
55 if not os.path.exists(args.o):
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
56 os.makedirs(args.o)
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
57 print('in file', args.i)
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
58 print('out dir', args.o)
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
59 print('nm files', args.n)
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
60
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
61 msp_split(args.i, args.o, int(args.n))
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
62
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
63
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
64 if __name__ == '__main__':
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
65 main()
bd298a4f21e4 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f79fa34772bbab836d89cf8bad52d49285409a98
tomnl
parents:
diff changeset
66