def extract_barcodes(fqfile,
lane,
nindex=25,
casava18=True,
offset=101,
bclen=6,
expected=[],
mismatch=True):
"""Parse the fastq file and extract barcodes. Return a dict structure suitable for upload to StatusDB
"""
bcx = BarcodeExtractor(fqfile, casava18, offset, bclen)
c = Counter(bcx)
c = remove_expected(c, expected, mismatch)
counts = []
header = ['lane', 'sequence', 'count', 'index_name']
for bc, count in c.most_common(nindex):
counts.append(
dict(
zip(header, [
lane, bc, count, ','.join(map_index_name(
bc, int(mismatch)))
])))
return [header, counts]
def test_map_index_name(self):
"""Map index sequences to names
"""
# Some test cases
from scilifelab.illumina.index_definitions import BASIC_LOOKUP
random_keys = [BASIC_LOOKUP.keys()[random.randint(0,len(BASIC_LOOKUP.keys())-1)] for n in xrange(25)]
for name in random_keys:
self.assertIn(name,map_index_name(BASIC_LOOKUP[name],0),
"Exact mapping did not return expected index name")
def extract_barcodes(fqfile, lane, nindex=25, casava18=True, offset=101, bclen=6, expected=[], mismatch=True):
"""Parse the fastq file and extract barcodes. Return a dict structure suitable for upload to StatusDB
"""
bcx = BarcodeExtractor(fqfile, casava18, offset, bclen)
c = Counter(bcx)
c = remove_expected(c,expected,mismatch)
counts = []
header = ['lane', 'sequence', 'count', 'index_name']
for bc, count in c.most_common(nindex):
counts.append(dict(zip(header,
[lane, bc, count, ','.join(map_index_name(bc,int(mismatch)))])))
return [header, counts]
def test_map_index_name(self):
"""Map index sequences to names
"""
# Some test cases
from scilifelab.illumina.index_definitions import BASIC_LOOKUP
random_keys = [
BASIC_LOOKUP.keys()[random.randint(0,
len(BASIC_LOOKUP.keys()) - 1)]
for n in xrange(25)
]
for name in random_keys:
self.assertIn(name, map_index_name(BASIC_LOOKUP[name], 0),
"Exact mapping did not return expected index name")
