def test_modern(self):
"""Testing IlluminaFastqRecord"""
fil = os.path.join(os.path.dirname(__file__),
'etc/fake_IlluminaSequenceIdentifier.fastq')
m = infer_fastq_type(fil)
expected = ('IlluminaFastqRecord', IlluminaFastqRecord)
self.assertEqual(expected, m)
def test_nobarcode(self):
"""Testing IlluminaNoBarcodeFastqRecord"""
fil = os.path.join(
os.path.dirname(__file__),
"etc/fake_IlluminaSequenceIdentifierNoBarcode.fastq",
)
m = infer_fastq_type(fil)
expected = ("IlluminaNoBarcodeFastqRecord",
IlluminaNoBarcodeFastqRecord)
self.assertEqual(expected, m)
def process_fastq(
input_file, output_prefix, logger_name="fastq_processing", log_itvl=1000000
):
"""
Processes the provided fastq file and splits into 1 or more separate
readgroup level fastq files.
:param input_file: input fastq file path
:param output_prefix: output prefix
:param loggger_name: name of the logger created
:param log_itvl: print log every N records
:return: a tuple containing dictionary of report and total counts
"""
logger = get_logger(logger_name)
ibase = os.path.basename(input_file)
logger.info("Processing fastq: {0}".format(ibase))
fq_cls = infer_fastq_type(input_file)
logger.info("Inferred fastq class {0} in fastq {1}".format(fq_cls[0], ibase))
reader = FastqReader(input_file, record_cls=fq_cls[1])
count = 0
writers = {}
try:
for record in reader:
count += 1
if count % log_itvl == 0:
logger.info("Processed {0} records from {1}".format(count, ibase))
key = record.read_key
if key not in writers:
logger.info("Found read key {0} in fastq {1}".format(key, ibase))
writer = FastqWriterWithReport.from_record_and_prefix(
record, output_prefix
)
logger.info(
"Output file for read key {0} in fastq {1} is {2}".format(
key, ibase, writer.fname
)
)
writers[key] = writer
writers[key] += record
finally:
reader.close()
for key in writers:
writers[key].close()
logger.info(
"Processed a total of {0} records from {1} and found {2} read keys".format(
count, ibase, len(writers)
)
)
return ({key: writers[key].reporter.to_dict() for key in writers}, count)
def test_unknown(self):
"""Testing raise exception"""
fil = os.path.join(os.path.dirname(__file__), 'etc/fake_Unknown.fastq')
with self.assertRaises(Exception):
m = infer_fastq_type(fil)