annotate galaxy-tools/biobank/updater/update_parents.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 import sys, csv, argparse, time, json
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
2
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
3 from bl.vl.kb import KnowledgeBase as KB
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
4 import bl.vl.utils.ome_utils as vlu
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
5 from bl.vl.utils import get_logger, LOG_LEVELS
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
6
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
7
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
8 def make_parser():
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
9 parser = argparse.ArgumentParser(description='update parents')
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
10 parser.add_argument('--logfile', type=str, help='log file (default=stderr)')
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
11 parser.add_argument('--loglevel', type=str, choices=LOG_LEVELS,
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
12 help='logging level (default=INFO)', default='INFO')
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
13 parser.add_argument('-H', '--host', type=str, help='omero hostname')
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
14 parser.add_argument('-U', '--user', type=str, help='omero user')
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
15 parser.add_argument('-P', '--passwd', type=str, help='omero password')
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
16 parser.add_argument('-O', '--operator', type=str, help='operator',
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
17 required=True)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
18 parser.add_argument('--in_file', type=str, required=True,
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
19 help='input file with individual, father and mother')
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
20 return parser
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
21
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
22
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
23 def update_parents(individual, father, mother, operator, kb, logger):
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
24 backup = {}
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
25 logger.info('Updating parents for individual %s', individual.id)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
26 if individual.father != father:
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
27 backup['father'] = individual.father.id if individual.father else None
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
28 logger.info('Setting father to %s (old value %s)' % (father.id if father else None,
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
29 backup['father']))
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
30 individual.father = father
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
31 if individual.mother != mother:
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
32 backup['mother'] = individual.mother.id if individual.mother else None
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
33 logger.info('Setting mother to %s (old value %s)' % (mother.id if mother else None,
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
34 backup['mother']))
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
35 individual.mother = mother
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
36 if len(backup.items()) > 0:
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
37 update_object(individual, backup, operator, kb, logger)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
38 return individual
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
39 else:
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
40 logger.info('No update needed for individual %s' % individual.id)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
41 return None
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
42
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
43
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
44 def update_object(obj, backup_values, operator, kb, logger):
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
45 logger.debug('Building ActionOnAction for object %s' % obj.id)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
46 act_setup = build_action_setup('update-parents-%f' % time.time(),
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
47 backup_values, kb, logger)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
48 aoa_conf = {
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
49 'setup': act_setup,
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
50 'actionCategory': kb.ActionCategory.UPDATE,
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
51 'operator': operator,
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
52 'target': obj.lastUpdate if obj.lastUpdate else obj.action,
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
53 'context': obj.action.context
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
54 }
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
55 logger.debug('Updating object with new ActionOnAction')
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
56 obj.lastUpdate = kb.factory.create(kb.ActionOnAction, aoa_conf)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
57
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
58
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
59 def build_action_setup(label, backup, kb, logger):
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
60 logger.debug('Creating a new ActionSetup with label %s and backup %r' % (label,
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
61 backup))
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
62 conf = {
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
63 'label': label,
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
64 'conf': json.dumps({'backup': backup})
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
65 }
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
66 asetup = kb.factory.create(kb.ActionSetup, conf)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
67 return asetup
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
68
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
69
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
70 def main(argv):
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
71 parser = make_parser()
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
72 args = parser.parse_args(argv)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
73
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
74 logger = get_logger('update_parents', level=args.loglevel,
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
75 filename=args.logfile)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
76
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
77 try:
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
78 host = args.host or vlu.ome_host()
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
79 user = args.user or vlu.ome_user()
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
80 passwd = args.passwd or vlu.ome_passwd()
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
81 except ValueError, ve:
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
82 logger.critical(ve)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
83 sys.exit(ve)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
84
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
85 kb = KB(driver='omero')(host, user, passwd)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
86
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
87 logger.info('Retrieving individuals')
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
88 inds = kb.get_objects(kb.Individual)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
89 logger.info('Retrieved %d individuals' % len(inds))
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
90 inds_lookup = {}
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
91 for i in inds:
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
92 inds_lookup[i.id] = i
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
93
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
94 with open(args.in_file) as in_file:
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
95 to_be_updated = []
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
96 reader = csv.DictReader(in_file, delimiter='\t')
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
97 for row in reader:
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
98 ind = inds_lookup[row['individual']]
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
99 father = inds_lookup[row['father']] if row['father'] != 'None' else None
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
100 mother = inds_lookup[row['mother']] if row['mother'] != 'None' else None
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
101 ind = update_parents(ind, father, mother, args.operator, kb, logger)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
102 if ind:
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
103 to_be_updated.append(ind)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
104
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
105 logger.info('%d individuals are going to be updated' % len(to_be_updated))
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
106 kb.save_array(to_be_updated)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
107 logger.info('Update complete')
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
108
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
109 if __name__ == '__main__':
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
110 main(sys.argv[1:])