annotate galaxy-tools/biobank/utils/delete_flowcell_results.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 argparse, sys
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 from bl.vl.utils import get_logger, LOG_LEVELS
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
5 import bl.vl.utils.ome_utils as vlu
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='remove datasamples connected to a specific sample of a flowcell')
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 server 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('--flowcell-label', type=str, required=True,
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
17 help='flowcell label')
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
18 parser.add_argument('--sample-label', type=str, required=True,
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
19 help='sample label')
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
20 parser.add_argument('--dry-run', action='store_true',
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
21 help='run a simulation, don\'t delete anything')
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
22 return parser
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
23
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
24
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
25 def get_flowcell_samples_map(flowcell, kb, logger):
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
26 fc = kb.get_by_label(kb.FlowCell, flowcell)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
27 if not fc:
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
28 logger.info('No flowcell with label %s', flowcell)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
29 sys.exit(0)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
30 logger.info('Loading data for flowcell %s', flowcell)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
31 dsamples = kb.dt.get_connected(fc, kb.SeqDataSample)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
32 dsamples_map = {}
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
33 for ds in dsamples:
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
34 if ds.sample:
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
35 dsamples_map.setdefault(ds.sample.label, []).append(ds)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
36 else:
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
37 dsamples_map.setdefault('NO_SAMPLE', []).append(ds)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
38 return dsamples_map
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
39
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
40
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
41 def print_report(dsamples_map, sample_label, kb, logger):
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
42 dsamples = dsamples_map.get(sample_label)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
43 if not dsamples:
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
44 logger.info('No sample with label %s is related to the flowcell', sample_label)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
45 sys.exit(0)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
46 for ds in dsamples:
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
47 dobjs = kb.get_data_objects(ds)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
48 logger.info('## data sample: %s', ds.label)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
49 for dob in dobjs:
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
50 logger.info('### data object: %s --- mimetype: %s', dob.path, dob.mimetype)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
51
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
52
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
53 def delete(dsamples_map, sample_label, kb, logger):
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
54 for ds in dsamples_map[sample_label]:
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
55 # this is a hack specific for the automator workflow
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
56 if not ds.label.startswith('stage1'):
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
57 logger.info('Deleting data for %s', ds.label)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
58 dobjs = kb.get_data_objects(ds)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
59 for d in dobjs:
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
60 kb.delete(d)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
61 a = ds.action
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
62 kb.delete(ds)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
63 try:
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
64 kb.delete(a)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
65 except:
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
66 pass
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
67
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
68
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
69 def main(argv):
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
70 parser = make_parser()
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
71 args = parser.parse_args(argv)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
72
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
73 logger = get_logger('delete_flowcell_results', level=args.loglevel,
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
74 filename=args.logfile)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
75
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
76 try:
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
77 host = args.host or vlu.ome_host()
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
78 user = args.user or vlu.ome_user()
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
79 passwd = args.passwd or vlu.ome_passwd()
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
80 except ValueError, ve:
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
81 logger.critical(ve)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
82 sys.exit(ve)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
83
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
84 kb = KB(driver='omero')(host, user, passwd)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
85
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
86 dsamples_map = get_flowcell_samples_map(args.flowcell_label, kb, logger)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
87 print_report(dsamples_map, args.sample_label, kb, logger)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
88 if not args.dry_run:
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
89 delete(dsamples_map, args.sample_label, kb. logger)
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
90 pass
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
91 else:
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
92 logger.debug('SIMULATION, exit now')
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
93 logger.info('Job completed')
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
94
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
95
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
96 if __name__ == '__main__':
ba6cf6ede027 Uploaded
ric
parents:
diff changeset
97 main(sys.argv[1:])