def __init__(self, args=None, logger=None):
self.logger = logger
self.output_format = args.output_format
self.conf = get_conf(logger, args.config_file)
self.io_conf = self.conf.get_io_section()
c = Client(conf=self.conf, logger=self.logger)
c.init_bika()
self.bika = c.bk
path_exists(args.input_file, self.logger)
with open(args.input_file, 'r') as stream:
input_data = yaml.safe_load(stream)
self.input_paths = input_data.get('default_paths',
[self.io_conf.get('archive_root_path')])
for _ in self.input_paths:
path_exists(_, self.logger)
output_file = args.output_file if args.output_file else None
if output_file != os.path.realpath(output_file):
self.logger.error('{} is not a valid path. Please use absolute path'.format(output_file))
sys.exit()
self.output_file = output_file
batches = input_data.get('batches', None)
self.batches_info = dict()
self.sids = list()
for _, batch in batches.items():
bid = batch.get('bid', None)
samples = batch.get('samples', [])
if bid:
self.logger.info("Retrieving info for batch {}".format(bid))
batch_info = self.bika.get_batch_info(bid, samples)
if batch_info:
sids = [_ for _ in batch_info.keys() if batch_info[
_].get('type') not in SAMPLE_TYPES_TOSKIP]
self.sids.extend(sids)
self.batches_info.update(batch_info)
else:
self.logger.error('No samples information found for the '
'batch {}'.format(bid))
if not self.sids:
self.logger.error('I have not retrieve any information for the '
'batches {}'.format(" ".join(self.sids)))
sys.exit()
def __init__(self, args=None, logger=None):
self.logger = logger
self.conf = get_conf(logger, args.config_file)
self.io_conf = self.conf.get_io_section()
self.do_conf = self.conf.get_section('data_ownership')
self.ir_conf = self.conf.get_irods_section()
self.root_path = args.root_path if args.root_path else self.io_conf.get('rundirs_root_path')
self.user = self.do_conf.get('user')
self.group = self.do_conf.get('group')
self.emit_events = args.emit_events
def __init__(self, args=None, logger=None):
self.logger = logger
conf = get_conf(logger, args.config_file)
self.conf = conf
self.irods_conf = self.conf.get_irods_section()
self.bika_conf = self.conf.get_section('bika')
self.io_conf = self.conf.get_io_section()
self.rundir_label = args.rundir_label
self.samplesheet_filename = 'SampleSheet.csv'
self.emit_events = args.emit_events
self.force = args.force
self.sync_all_analyses = args.sync_all_analyses
def __init__(self, args=None, logger=None):
self.logger = logger
conf = get_conf(logger, args.config_file)
self.conf = conf
self.irods_conf = self.conf.get_irods_section()
self.bika_conf = self.conf.get_section('bika')
self.io_conf = self.conf.get_io_section()
self.rundir_label = args.rundir_label
self.samplesheet_filename = 'SampleSheet.csv'
self.emit_events = args.emit_events
self.force = args.force
self.sync_all_analyses = args.sync_all_analyses
def __init__(self, args=None, logger=None):
self.logger = logger
conf = get_conf(logger, args.config_file)
if args.root_path:
self.root_path = args.root_path
else:
io_conf = conf.get_io_section()
self.root_path = io_conf.get('rundirs_root_path')
do_conf = conf.get_section('data_ownership')
self.user = do_conf.get('user')
self.group = do_conf.get('group')
self.ir_conf = conf.get_irods_section()
self.proc_rundir = args.proc_rundir
def __init__(self, args=None, logger=None):
self.logger = logger
conf = get_conf(logger, args.config_file)
self.batch_queuing = args.batch_queuing
self.queues_conf = conf.get_section('queues')
r_dir_label = args.rundir_label
ds_dir_label = 'datasets'
fqc_dir_label = 'fastqc'
input_path = args.ds_path
output_path = args.export_path
if r_dir_label or (input_path and output_path):
pass
else:
logger.error("You must provide the rundir_label or both ds_path "
"and export_path")
sys.exit()
# input path must exists as parser argument or as config file argument
if not input_path:
io_conf = conf.get_io_section()
input_path = os.path.join(io_conf.get('archive_root_path'),
r_dir_label, ds_dir_label)
path_exists(input_path, logger)
self.input_path = input_path
# export path must exists as parser argument or as config file argument
if not output_path:
io_conf = conf.get_io_section()
output_path = os.path.join(io_conf.get('qc_export_basepath'),
r_dir_label)
# FIXME: this is a local path, must be checked that run on right node
if not path_exists(output_path, logger, force=False):
ensure_dir(output_path)
path_exists(output_path, logger)
self.output_path = output_path
self.fqc_path = os.path.join(self.input_path, fqc_dir_label)
def __init__(self, args=None, logger=None):
self.logger = logger
self.destination = args.destination
self.dry_run = args.dry_run
conf = get_conf(logger, args.config_file)
self.conf = conf
self.batch_id = batch_id = args.batch_id
c = Client(conf=conf, logger=logger)
c.init_bika()
batch_info = c.bk.get_batch_info(batch_id)
if batch_info:
self.batch_info = batch_info
else:
logger.error('I have not found any information of the samples '
'owned by the batch {}'.format(batch_id))
sys.exit()
# input path must exists as parser argument or as config file argument
if args.input_path:
input_path = args.input_path
else:
io_conf = conf.get_io_section()
input_path = io_conf.get('archive_root_path')
path_exists(input_path, logger)
self.input_path = input_path
output_path = args.output_path if args.output_path else None
self.output_path = output_path
inventory = args.inventory if args.inventory else None
self.inventory = inventory
playbook_path = args.playbook_path if args.playbook_path else None
self.playbook_path = playbook_path
def __init__(self, args=None, logger=None):
self.logger = logger
self.conf = get_conf(logger, args.config_file)
self.io_conf = self.conf.get_io_section()
self.batch_queuing = args.batch_queuing
self.queues_conf = self.conf.get_section('queues')
rd_label = args.rd_label
ds_path = args.ds_path if args.ds_path \
else os.path.join(self.io_conf.get('archive_root_path'),
rd_label,
self.io_conf.get('ds_folder_name'))
qc_path = args.qc_path if args.qc_path \
else os.path.join(ds_path,
self.io_conf.get('qc_folder_name'))
qc_export_path = args.qc_export_path if args.qc_export_path \
else os.path.join(self.io_conf.get('qc_export_basepath'),
rd_label)
# FIXME: this is a local path, must be checked that run on right node
if not path_exists(qc_export_path, logger, force=False):
ensure_dir(qc_export_path)
path_exists(ds_path, logger)
path_exists(qc_export_path, logger)
self.input_path = ds_path
self.output_path = qc_export_path
self.qc_path = qc_path
self.rerun = args.rerun
self.started = os.path.join(self.qc_path,
self.io_conf.get('quality_check_started_file'))
self.completed = os.path.join(self.qc_path,
self.io_conf.get('quality_check_completed_file'))
def __init__(self, args=None, logger=None):
self.logger = logger
self.conf = get_conf(logger, args.config_file)
self.io_conf = self.conf.get_io_section()
self.batch_queuing = args.batch_queuing
self.queues_conf = self.conf.get_section('queues')
rd_label = args.rd_label
ds_path = args.ds_path if args.ds_path \
else os.path.join(self.io_conf.get('archive_root_path'),
rd_label,
self.io_conf.get('ds_folder_name'))
qc_path = args.qc_path if args.qc_path \
else os.path.join(ds_path,
self.io_conf.get('qc_folder_name'))
qc_export_path = args.qc_export_path if args.qc_export_path \
else os.path.join(self.io_conf.get('qc_export_basepath'),
rd_label)
# FIXME: this is a local path, must be checked that run on right node
if not path_exists(qc_export_path, logger, force=False):
ensure_dir(qc_export_path)
path_exists(ds_path, logger)
path_exists(qc_export_path, logger)
self.input_path = ds_path
self.output_path = qc_export_path
self.qc_path = qc_path
self.rerun = args.rerun
self.started = os.path.join(
self.qc_path, self.io_conf.get('quality_check_started_file'))
self.completed = os.path.join(
self.qc_path, self.io_conf.get('quality_check_completed_file'))
def __init__(self, args=None, logger=None):
self.logger = logger
rpath = args.rd_path
cpath = args.rd_path.replace('running', 'completed')
apath = os.path.join(cpath, 'raw')
self.rd = {
'rpath': rpath,
'cpath': cpath,
'apath': apath,
'label': os.path.basename(args.rd_path)
}
conf = get_conf(logger, args.config_file)
self.conf = conf
dspath = os.path.join(cpath, 'datasets')
self.ds = {'path': dspath}
fqc_basepath = os.path.join(dspath, 'fastqc')
self.fqc = dict(path=fqc_basepath)
io_conf = conf.get_io_section()
export_path = os.path.join(io_conf.get('qc_export_basepath'),
self.rd['label'])
self.fqc.update(dict(export_path=export_path))
ssheet = {
'basepath': os.path.join(cpath),
'filename': 'SampleSheet.csv'
}
ssheet['file_path'] = os.path.join(ssheet['basepath'],
ssheet['filename'])
self.samplesheet = ssheet
run_info = {'basepath': os.path.join(rpath), 'filename': 'RunInfo.xml'}
run_info['file_path'] = os.path.join(run_info['basepath'],
run_info['filename'])
run_info['file_apath'] = os.path.join(apath, run_info['filename'])
self.run_info = run_info
run_parameters = {
'basepath': os.path.join(rpath),
'filename': 'runParameters.xml'
}
run_parameters['file_path'] = os.path.join(run_parameters['basepath'],
run_parameters['filename'])
run_parameters['file_apath'] = os.path.join(apath,
run_parameters['filename'])
self.run_parameters = run_parameters
do_conf = conf.get_section('data_ownership')
self.user = do_conf.get('user')
self.group = do_conf.get('group')
self.no_lane_splitting = args.no_lane_splitting
self.barcode_mismatches = args.barcode_mismatches
self.overwrite_samplesheet = args.overwrite_samplesheet
self.copy_qc = args.export_qc
self.batch_queuing = args.batch_queuing
self.queues_conf = conf.get_section('queues')
self._add_config_from_cli(args)
def __init__(self, args=None, logger=None):
self.logger = logger
conf = get_conf(logger, args.config_file)
self.conf = conf
self.io_conf = self.conf.get_io_section()
self.ir_conf = self.conf.get_irods_section()
self.do_conf = self.conf.get_section('data_ownership')
self.queues_conf = self.conf.get_section('queues')
rd_label = args.rd_label
rd_path = args.rd_path if args.rd_path \
else os.path.join(self.io_conf.get('rundirs_root_path'),
rd_label)
ds_path = args.ds_path if args.ds_path \
else os.path.join(self.io_conf.get('archive_root_path'),
rd_label,
self.io_conf.get('ds_folder_name'))
ssheet_path = args.ssheet_path if args.ssheet_path \
else os.path.join(self.io_conf.get('archive_root_path'),
rd_label,
self.io_conf.get('ssheet_filename'))
qc_path = args.qc_path if args.qc_path \
else os.path.join(ds_path,
self.io_conf.get('qc_folder_name'))
qc_export_path = args.qc_export_path if args.qc_export_path \
else os.path.join(self.io_conf.get('qc_export_basepath'),
rd_label)
logbook_path = args.logbook if args.logbook \
else os.path.join(ds_path,
self.io_conf.get('logbook_filename'))
run_info_path = os.path.join(
rd_path,
self.io_conf.get('run_info_filename')
)
run_parameters_path = os.path.join(
rd_path,
self.io_conf.get('run_parameters_filename')
)
self.rd = dict(
path=rd_path,
label=rd_label
)
self.ds = dict(
path=ds_path
)
self.qc = dict(
path=qc_path,
export_path=qc_export_path
)
self.samplesheet = dict(
path=ssheet_path,
filename=os.path.basename(ssheet_path)
)
self.logbook = dict(
path=logbook_path,
filename=os.path.basename(logbook_path)
)
self.run_info = dict(
path=run_info_path,
filename=os.path.basename(run_info_path)
)
self.run_parameters = dict(
path=run_parameters_path,
filename=os.path.basename(run_parameters_path)
)
self.user = self.do_conf.get('user')
self.group = self.do_conf.get('group')
self.no_lane_splitting = args.no_lane_splitting
self.barcode_mismatches = args.barcode_mismatches
self.with_failed_reads = args.with_failed_reads
self.overwrite_samplesheet = not os.path.isfile(ssheet_path)
self.emit_events = args.emit_events
self.batch_queuing = args.batch_queuing
self.started_file = os.path.join(self.rd['path'],
self.io_conf.get('preprocessing_started_file'))
self.completed_file = os.path.join(self.rd['path'],
self.io_conf.get('preprocessing_completed_file'))