diff galaxy-tools/biobank/utils/drop_flowcell_related_items.py @ 3:43be74e62bfe draft

Uploaded
author ric
date Thu, 22 Sep 2016 08:57:04 -0400
parents
children
line wrap: on
line diff
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/galaxy-tools/biobank/utils/drop_flowcell_related_items.py	Thu Sep 22 08:57:04 2016 -0400
@@ -0,0 +1,124 @@
+import argparse, sys
+from collections import Counter
+
+from bl.vl.kb import KnowledgeBase as KB
+import bl.vl.utils.ome_utils as vlu
+from bl.vl.utils import get_logger, LOG_LEVELS
+from bl.vl.kb.drivers.omero.sequencing import SeqDataSample, SequencerOutput
+
+
+def make_parser():
+    parser = argparse.ArgumentParser(description='delete all items related to the given flowcell')
+    parser.add_argument('--logfile', type=str, help='log file (default=stderr)')
+    parser.add_argument('--loglevel', type=str, choices=LOG_LEVELS,
+                        help='logging level (default=INFO)', default='INFO')
+    parser.add_argument('-H', '--host', type=str, help='omero hostname')
+    parser.add_argument('-U', '--user', type=str, help='omero user')
+    parser.add_argument('-P', '--passwd', type=str, help='omero password')
+    parser.add_argument('--flowcell-label', type=str, required=True,
+                        help='flowcell label')
+    parser.add_argument('--datasets-only', action='store_true',
+                        help='delete only data samples and data objects related to the given flowcell')
+    return parser
+
+
+def get_sources(objs):
+    sources = set()
+    for x in objs:
+        try:
+            sources.add(x.action.target)
+        except AttributeError:
+            # Action has no "target" attribute, no source for item x
+            pass
+    return sources
+
+
+def delete_objects(objs, kb, logger, max_retries = 3):
+    retry_ct = Counter()
+    while len(objs) > 0:
+        o = objs.pop(0)
+        if type(o) in [SeqDataSample, SequencerOutput]:
+            logger.info('Loading DataObjects for %s:%s' % (o.__class__.__name__,
+                                                           o.label))
+            dobjs = kb.get_data_objects(o)
+            logger.info('%d DataObjects loaded' % len(dobjs))
+            for d in dobjs:
+                logger.info('Deleting %s:%s' % (d.__class__.__name__,
+                                                d.path))
+                kb.delete(d)
+        try:
+            logger.info('Deleting %s:%s' % (o.__class__.__name__,
+                                            o.id))
+            act = o.action
+            kb.delete(o)
+            try:
+                logger.info('Deleting source action %s:%s' % (act.__class__.__name__,
+                                                              act.id))
+                kb.delete(act)
+            except:
+                logger.info('Can\'t delete action')
+        except:
+            logger.info('Can\'t delete, putting back into objects list')
+            if retry_ct['%s:%s' % (type(o), o.id)] < max_retries:
+                objs.append(o)
+                retry_ct['%s:%s' % (type(o), o.id)] += 1
+            else:
+                logger.info('Reached maximum retry limit for the object, skipping')
+
+def main(argv):
+    parser = make_parser()
+    args = parser.parse_args(argv)
+
+    logger = get_logger('drop_flowcell_related_items', level=args.loglevel,
+                        filename=args.logfile)
+
+    try:
+        host = args.host or vlu.ome_host()
+        user = args.user or vlu.ome_user()
+        passwd = args.passwd or vlu.ome_passwd()
+    except ValueError, ve:
+        logger.critical(ve)
+        sys.exit(ve)
+
+    kb = KB(driver='omero')(host, user, passwd)
+
+    logger.info('Retrieving flowcell with label %s' % args.flowcell_label)
+    query = 'SELECT fc FROM FlowCell fc WHERE fc.label = :fc_label'
+    results = kb.find_all_by_query(query, {'fc_label' : args.flowcell_label})
+    if len(results) == 0:
+        logger.info('No FlowCell with label %s in the database' % args.flowcell_label)
+        sys.exit(0)
+    fc = results[0]
+    logger.info('Loaded FlowCell with ID %s' % fc.id)
+    logger.info('Loading related Lanes')
+    lanes = list(kb.get_lanes_by_flowcell(fc))
+    logger.info('%d Lanes loaded' % len(lanes))
+    logger.info('Loading related LaneSlots')
+    lane_slots = []
+    for l in lanes:
+        lane_slots += list(kb.get_laneslots_by_lane(l))
+    logger.info('%d LaneSlots loaded' % len(lane_slots))
+    logger.info('Loading related Tubes')
+    sub_samples = get_sources(lane_slots)
+    samples = get_sources(sub_samples)
+    logger.info('%d Tubes loaded' % (len(sub_samples) + len(samples)))
+    logger.info('Loading related SequencerOutputs')
+    seq_out = kb.dt.get_connected(fc, aklass = kb.SequencerOutput, direction = kb.dt.DIRECTION_OUTGOING)
+    logger.info('%d SequencerOutputs loaded' % len(seq_out))
+    logger.info('Loading related SeqDataSamples')
+    seq_dsamples = kb.dt.get_connected(fc, aklass = kb.SeqDataSample, direction = kb.dt.DIRECTION_OUTGOING)
+    logger.info('%d SeqDataSamples loaded' % len(seq_dsamples))
+
+    if args.datasets_only:
+        delete_items = [seq_dsamples]
+    else:
+        delete_items = [seq_dsamples, seq_out, lane_slots, lanes,
+                        [fc], list(sub_samples), list(samples)]
+    for items in delete_items:
+        delete_objects(items, kb, logger)
+
+    logger.info('Job completed')
+
+
+if __name__ == '__main__':
+    main(sys.argv[1:])