def test_iterate_fastq_reader(self, fastq_filename):
fastq_path = test_utils.genomics_core_testdata(fastq_filename)
expected_ids = [
'NODESC:header', 'M01321:49:000000000-A6HWP:1:1101:17009:2216',
'FASTQ'
]
with fastq.FastqReader(fastq_path) as reader:
records = list(reader.iterate())
self.assertLen(records, 3)
self.assertEqual([r.id for r in records], expected_ids)
def test_roundtrip_writer(self, filename):
output_path = test_utils.test_tmpfile(filename)
with fastq.FastqWriter(output_path) as writer:
for record in self.records:
writer.write(record)
with fastq.FastqReader(output_path) as reader:
v2_records = list(reader.iterate())
self.assertEqual(self.records, v2_records)
def test_round_trip_fastq(self, test_datum_name):
# Round-trip FASTQ records through writing and reading:
# 1. Read records v1 from FastqReader;
# 2. Write v1 to fastq using our FastqWriter;
# 3. Read back in using FastqReader -- v2;
# 4. compare v1 and v2.
in_file = test_utils.genomics_core_testdata(test_datum_name)
out_file = test_utils.test_tmpfile('output_' + test_datum_name)
v1_reader = fastq.FastqReader(in_file)
v1_records = list(v1_reader.iterate())
self.assertTrue(v1_records, 'Reader failed to find records')
writer_options = fastq_pb2.FastqWriterOptions()
with fastq_writer.FastqWriter.to_file(out_file, writer_options) as writer:
for record in v1_records:
writer.write(record)
v2_reader = fastq.FastqReader(out_file)
v2_records = list(v2_reader.iterate())
self.assertEqual(v1_records, v2_records,
'Round-tripped FASTQ files not as expected')