def test_fastq_iterate(self):
with fastq_reader.FastqReader.from_file(self.fastq,
self.options) as reader:
iterable = reader.iterate()
self.assertIsInstance(iterable, clif_postproc.WrappedCppIterable)
self.assertEqual(test_utils.iterable_len(iterable), 3)
zreader = fastq_reader.FastqReader.from_file(
self.zipped_fastq,
fastq_pb2.FastqReaderOptions(
compression_type=fastq_pb2.FastqReaderOptions.GZIP))
with zreader:
ziterable = zreader.iterate()
self.assertIsInstance(ziterable, clif_postproc.WrappedCppIterable)
self.assertEqual(test_utils.iterable_len(ziterable), 3)
def test_bam_iterate(self):
reader = sam_reader.SamReader.from_file(reads_path=self.bam,
ref_path='',
options=self.options)
with reader:
iterable = reader.iterate()
self.assertIsInstance(iterable, clif_postproc.WrappedCppIterable)
self.assertEqual(test_utils.iterable_len(iterable), 106)
def test_sam_query(self):
reader = sam.SamReader(test_utils.genomics_core_testdata('test.bam'))
expected = [(ranges.parse_literal('chr20:10,000,000-10,000,100'), 106),
(ranges.parse_literal('chr20:10,000,000-10,000,000'), 45)]
with reader:
for interval, n_expected in expected:
with reader.query(interval) as iterable:
self.assertEqual(test_utils.iterable_len(iterable), n_expected)
def test_bam_query(self):
reader = sam_reader.SamReader.from_file(self.bam, self.indexed_options)
expected = [(ranges.parse_literal('chr20:10,000,000-10,000,100'), 106),
(ranges.parse_literal('chr20:10,000,000-10,000,000'), 45)]
with reader:
for interval, n_expected in expected:
with reader.query(interval) as iterable:
self.assertIsInstance(iterable, clif_postproc.WrappedCppIterable)
self.assertEqual(test_utils.iterable_len(iterable), n_expected)
def test_downsampling(self, method, maybe_range, fraction, expected_n_reads):
reader = sam.SamReader(
test_utils.genomics_core_testdata('test.bam'),
downsample_fraction=fraction,
random_seed=12345)
with reader:
if method == 'iterate':
reads_iter = reader.iterate()
elif method == 'query':
reads_iter = reader.query(ranges.parse_literal(maybe_range))
else:
self.fail('Unexpected method ' + str(method))
self.assertEqual(test_utils.iterable_len(reads_iter), expected_n_reads)
def test_vcf_query(self):
range1 = ranges.parse_literal('chr3:100,000-500,000')
iterable = self.samples_reader.query(range1)
self.assertEqual(test_utils.iterable_len(iterable), 4)
def test_vcf_iterate(self):
iterable = self.sites_reader.iterate()
self.assertEqual(test_utils.iterable_len(iterable), 5)
def test_fastq_iterate(self, filename):
path = test_utils.genomics_core_testdata(filename)
with fastq_reader.FastqReader.from_file(path, self.options) as reader:
iterable = reader.iterate()
self.assertIsInstance(iterable, clif_postproc.WrappedCppIterable)
self.assertEqual(test_utils.iterable_len(iterable), 4)
def test_bam_iterate(self):
reader = sam.SamReader(test_utils.genomics_core_testdata('test.bam'))
with reader:
self.assertEqual(test_utils.iterable_len(reader.iterate()), 106)
def test_query(self, filename, has_embedded_ref):
with self._make_reader(filename, has_embedded_ref) as reader:
for interval, n_expected in [('chr1:1-100', 3), ('chr2:1-121', 0)]:
with reader.query(ranges.parse_literal(interval)) as iterable:
self.assertEqual(test_utils.iterable_len(iterable),
n_expected)
def test_sam_iterate(self):
reader = sam.SamReader(test_utils.genomics_core_testdata('test.sam'),
use_index=False)
with reader:
self.assertEqual(test_utils.iterable_len(reader.iterate()), 6)
