def _collect_casava_qc(self):
qc_objects = []
runinfo_csv = os.path.join(os.path.abspath(self.pargs.flowcell), "{}.csv".format(self._fc_id()))
try:
with open(runinfo_csv) as fh:
runinfo_reader = csv.reader(fh)
runinfo = [x for x in runinfo_reader]
except IOError as e:
self.app.log.warn(str(e))
raise e
fcdir = os.path.join(os.path.abspath(self.pargs.analysis), self.pargs.flowcell)
(fc_date, fc_name) = self._fc_parts()
## Check modification time
if modified_within_days(fcdir, self.pargs.mtime):
fc_kw = dict(path=fcdir, fc_date = fc_date, fc_name=fc_name)
fcobj = FlowcellRunMetrics(**fc_kw)
fcobj.parse_illumina_metrics(fullRTA=False)
fcobj.parse_bc_metrics()
fcobj.parse_demultiplex_stats_htm()
fcobj.parse_samplesheet_csv()
qc_objects.append(fcobj)
for sample in runinfo[1:]:
d = dict(zip(runinfo[0], sample))
if self.app.pargs.project and self.app.pargs.project != d['SampleProject']:
continue
if self.app.pargs.sample and self.app.pargs.sample != d['SampleID']:
continue
sampledir = os.path.join(os.path.abspath(self.pargs.analysis), d['SampleProject'].replace("__", "."), d['SampleID'])
if not os.path.exists(sampledir):
self.app.log.warn("No such sample directory: {}".format(sampledir))
continue
sample_fcdir = os.path.join(sampledir, self._fc_fullname())
if not os.path.exists(sample_fcdir):
self.app.log.warn("No such sample flowcell directory: {}".format(sample_fcdir))
continue
if not modified_within_days(sample_fcdir, self.pargs.mtime):
continue
runinfo_yaml_file = os.path.join(sample_fcdir, "{}-bcbb-config.yaml".format(d['SampleID']))
if not os.path.exists(runinfo_yaml_file):
self.app.log.warn("No such yaml file for sample: {}".format(runinfo_yaml_file))
raise IOError(2, "No such yaml file for sample: {}".format(runinfo_yaml_file), runinfo_yaml_file)
with open(runinfo_yaml_file) as fh:
runinfo_yaml = yaml.load(fh)
if not runinfo_yaml['details'][0].get("multiplex", None):
self.app.log.warn("No multiplex information for sample {}".format(d['SampleID']))
continue
sample_kw = dict(path=sample_fcdir, flowcell=fc_name, date=fc_date, lane=d['Lane'], barcode_name=d['SampleID'], sample_prj=d['SampleProject'].replace("__", "."), barcode_id=runinfo_yaml['details'][0]['multiplex'][0]['barcode_id'], sequence=runinfo_yaml['details'][0]['multiplex'][0]['sequence'])
obj = SampleRunMetrics(**sample_kw)
obj.read_picard_metrics()
obj.parse_fastq_screen()
obj.parse_bc_metrics()
obj.read_fastqc_metrics()
qc_objects.append(obj)
return qc_objects
class TestQCUpload(unittest.TestCase):
def setUp(self):
if not os.path.exists(os.path.join(os.getenv("HOME"), "dbcon.ini")):
self.url = None
self.user = None
self.pw = None
self.examples = {}
else:
config = ConfigParser.ConfigParser()
config.readfp(open(os.path.join(os.getenv("HOME"), "dbcon.ini")))
self.url = config.get("couchdb", "url")
self.user = config.get("couchdb", "username")
self.pw = config.get("couchdb", "password")
self.demuxstats = config.get("statusdb", "demuxstats")
self.sample_kw = {
"path": config.get("statusdb", "fcdir"),
"flowcell": config.get("statusdb", "fc_name"),
"date": config.get("statusdb", "date"),
"lane": config.get("statusdb", "lane"),
"barcode_name": config.get("statusdb", "name"),
"sample_prj": config.get("statusdb", "project"),
"barcode_id": config.get("statusdb", "barcode_id"),
"sequence": config.get("statusdb", "sequence")
}
self.fc_kw = {
"path": config.get("statusdb", "fcdir"),
"fc_date": config.get("statusdb", "date"),
"fc_name": config.get("statusdb", "fc_name")
}
self.examples = {
"sample": config.get("examples", "sample"),
"flowcell": config.get("examples", "flowcell"),
"project": config.get("examples", "project")
}
self.fcrm = FlowcellRunMetrics(**self.fc_kw)
self.srm = SampleRunMetrics(**self.sample_kw)
def test_1_demuxstats(self):
metrics = self.fcrm.parse_demultiplex_stats_htm()
print metrics["Barcode_lane_statistics"][0]
def test_2_map_srmseqid_to_srmid(self):
"""Map srm seq id names to srm ids"""
sample_con = SampleRunMetricsConnection(username=self.user,
password=self.pw,
url=self.url)
sample_map = {}
for k in sample_con.db:
obj = sample_con.db.get(k)
sample_seq_id = "{}_{}_{}_{}".format(
obj.get("lane"), obj.get("date"), obj.get("flowcell"),
obj.get("sequence", "NoIndex"))
if not sample_seq_id in sample_map.keys():
sample_map[sample_seq_id] = [k]
else:
print "WARNING: duplicate for {}".format(sample_seq_id)
sample_map[sample_seq_id].append(k)
for k, v in sample_map.items():
if len(v) > 1:
print k, v
