annotate galaxy-tools/biobank/tools/get_parents_from_sibling.py @ 0:ba6cf6ede027 draft default tip

Uploaded
author ric
date Wed, 28 Sep 2016 06:03:30 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
1 '''
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
2 From a file like this
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
3 individual sibling
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
4 V08E18411BC66F4987BCA43EFC6F636224 V0AE5660BF4A7149589BE9DB3308B50327
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
5 V0FAE2B10F690041509739A3F4B314DC8F V00875417B31684EC2A62EE37717913445
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
6 V0382EF862AA4B475697C95D3777043239 V08E376727ED8E4B369DAA3B62A9395E1B
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
7 ....
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
8
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
9 retrieve indivual's parents using sibling informations and build a file like
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
10
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
11 individual father mother
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
12 V08E18411BC66F4987BCA43EFC6F636224 V027DE334753424F07B81A70053EF5B873 V035222CAEE0474AFEBB9A161D4B64914E
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
13 V0FAE2B10F690041509739A3F4B314DC8F V0E966B53BDCC942C09D6B6D96DE98F4F4 V0F7B6926C6FBE4F0BB38BBC6CFB13A825
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
14 ....
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
15
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
16 '''
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
17
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
18 import sys, csv, argparse, logging, os
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
19
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
20 from bl.vl.kb import KnowledgeBase as KB
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
21 import bl.vl.utils.ome_utils as vlu
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
22 from bl.vl.utils import get_logger, LOG_LEVELS
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
23
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
24
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
25 def make_parser():
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
26 parser = argparse.ArgumentParser(description='retrieve parents information using sibling')
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
27 parser.add_argument('--logfile', type=str, help='log file (default=stderr)')
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
28 parser.add_argument('--loglevel', type=str, choices=LOG_LEVELS,
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
29 help='logging level (default=INFO)', default='INFO')
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
30 parser.add_argument('-H', '--host', type=str, help='omero hostname')
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
31 parser.add_argument('-U', '--user', type=str, help='omero user')
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
32 parser.add_argument('-P', '--passwd', type=str, help='omero passwd')
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
33 parser.add_argument('--in_file', type=str, required=True,
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
34 help='input file with individual-sibling couples')
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
35 parser.add_argument('--out_file', type=str, required=True,
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
36 help='output file with parents information')
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
37 return parser
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
38
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
39
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
40 def main(argv):
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
41 parser = make_parser()
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
42 args = parser.parse_args(argv)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
43
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
44 logger = get_logger('get_parents_from_sibling', level=args.loglevel,
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
45 filename=args.logfile)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
46
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
47 try:
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
48 host = args.host or vlu.ome_host()
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
49 user = args.user or vlu.ome_user()
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
50 passwd = args.passwd or vlu.ome_passwd()
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
51 except ValueError, ve:
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
52 logger.critical(ve)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
53 sys.exit(ve)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
54
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
55 kb = KB(driver='omero')(host, user, passwd)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
56
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
57 logger.info('Retrieving individuals')
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
58 inds = kb.get_objects(kb.Individual)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
59 logger.info('Retrieved %d individuals' % len(inds))
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
60 inds_lookup = {}
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
61 for ind in inds:
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
62 inds_lookup[ind.id] = ind
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
63
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
64 with open(args.in_file) as infile:
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
65 reader = csv.DictReader(infile, delimiter='\t')
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
66 records = []
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
67 for row in reader:
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
68 try:
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
69 sib = inds_lookup[row['sibling']]
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
70 rec = {'individual' : row['individual'],
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
71 'father' : sib.father.id if sib.father else 'None',
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
72 'mother' : sib.mother.id if sib.mother else 'None'}
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
73 logger.info('Individual %s, father: %s - mother: %s' % (row['individual'],
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
74 rec['father'],
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
75 rec['mother']))
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
76 records.append(rec)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
77 except KeyError:
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
78 logger.error('Unable to find individual %s' % row['sibling'])
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
79
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
80 logger.info('Retrieved parents for %d individuals' % len(records))
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
81
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
82 with open(args.out_file, 'w') as outfile:
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
83 writer = csv.DictWriter(outfile, ['individual', 'father', 'mother'],
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
84 delimiter='\t')
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
85 writer.writeheader()
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
86 writer.writerows(records)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
87
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
88 if __name__ == '__main__':
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
89 main(sys.argv[1:])