view galaxy-tools/biobank/utils/from_well_to_illumina_measures.py @ 3:43be74e62bfe draft

Uploaded

import sys, argparse, csv

from bl.vl.kb import KnowledgeBase as KB
from bl.vl.utils import get_logger, LOG_LEVELS
import bl.vl.utils.ome_utils as vlu
from bl.vl.graph.drivers.neo4j import Neo4JDriver


def make_parser():
    parser = argparse.ArgumentParser(description='map wells label to illumina bead chip measures')
    parser.add_argument('--logfile', type=str, help='log file (deafult=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 server hostname')
    parser.add_argument('-U', '--user', type=str, help='OMERO user')
    parser.add_argument('-P', '--passwd', type=str, help='OMERO password')
    parser.add_argument('--in_file', type=str, help='input file',
                        required=True)
    parser.add_argument('--out_file', type=str, help='output file',
                        required=True)
    parser.add_argument('--well_column', type=str, help='label of the column that will be mapped',
                        default='source')
    return parser


def get_wells_map(plate, kb, logger):
    logger.info('Mapping wells for plate %s', plate.label)
    wells_map = {}
    for w in kb.get_wells_by_plate(plate):
        wells_map[w.label] = w
    logger.info('Mapped %d wells', len(wells_map))
    return wells_map


def get_plates_map(plates_list, kb, logger):
    logger.info('Loading TiterPlates map')
    plates_map = {}
    for pl in kb.get_objects(kb.TiterPlate):
        if isinstance(pl, kb.TiterPlate) and pl.barcode in plates_list:
            plates_map[pl.barcode] = get_wells_map(pl, kb, logger)
    logger.info('Mapped %d plates', len(plates_map))
    return plates_map


def get_connected_illumina_measures(well, kb, logger):
    logger.debug('Loading connected IlluminaBeadChipMeasures for well %s:%s', well.label,
                 well.container.label)
    return kb.dt.get_connected(well, aklass = kb.IlluminaBeadChipMeasures,
                               direction = Neo4JDriver.DIRECTION_OUTGOING)


def wells_to_illumina(in_file, out_file, column_label, kb, logger):
    with open(in_file) as ifile, open(out_file, 'w') as ofile:
        reader = csv.DictReader(ifile, delimiter='\t')
        in_records = [r for r in reader]
        plates_barcodes = set([x[column_label].split(':')[0] for x in in_records])
        plates_map = get_plates_map(plates_barcodes, kb, logger)
        writer = csv.DictWriter(ofile, reader.fieldnames, delimiter='\t')
        writer.writeheader()
        logger.info('Mapping wells to illumina bead chip measures')
        for rec in in_records:
            barcode, well = rec[column_label].split(':')
            measures = get_connected_illumina_measures(plates_map[barcode][well], kb,
                                                       logger)
            if len(measures) != 1:
                logger.warning('Found %d measures for well %s:%s, skipping line', len(measures),
                               barcode, well)
                continue
            rec[column_label] = measures[0].label
            writer.writerow(rec)


def main(argv):
    parser = make_parser()
    args = parser.parse_args(argv)

    logger = get_logger('from_well_to_illumina_measures', 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)
    wells_to_illumina(args.in_file, args.out_file, args.well_column,
                      kb, logger)
    logger.info('Job completed')


if __name__ == '__main__':
    main(sys.argv[1:])