def setUp(self):
self.hi = HiRapiAligner('rapi_bwa')
self.reads = test_utils.get_mini_ref_seqs()
for row in self.reads:
if len(row) != 5:
raise RuntimeError("Unexpected number of fields in mini_ref read record")
self.hi.load_pair(*row)
def _align_mini_ref_seqs(self):
self.hi.load_ref(test_utils.MiniRefMemPath)
reads = test_utils.get_mini_ref_seqs()
for row in reads:
if len(row) != 5:
raise RuntimeError("Unexpected number of fields in mini_ref read record")
self.hi.load_pair(*row)
self.hi.align_batch()
def setUp(self):
self.hi = HiRapiAligner('rapi_bwa')
self.reads = test_utils.get_mini_ref_seqs()
for row in self.reads:
if len(row) != 5:
raise RuntimeError(
"Unexpected number of fields in mini_ref read record")
self.hi.load_pair(*row)
def _align_mini_ref_seqs(self):
self.hi.load_ref(test_utils.MiniRefMemPath)
reads = test_utils.get_mini_ref_seqs()
for row in reads:
if len(row) != 5:
raise RuntimeError(
"Unexpected number of fields in mini_ref read record")
self.hi.load_pair(*row)
self.hi.align_batch()
def test_simple_map(self):
# get input data and expected output. We keep exactly one map batch of reads
# (as per self._map_batch_size)
reads = tseal_utils.get_mini_ref_seqs()[0:(self._map_batch_size / 2)]
expected_output = sorted(tseal_utils.rapi_mini_ref_seqs_sam_no_header().split('\n')[0:2*len(reads)])
self._log.info("loaded %s fragments and %s lines of expected output", len(reads), len(expected_output))
if len(reads) * 2 < self._mapper.batch_size:
self.fail("batch size for test (%s) is set larger than the number of available "
"reads (%s). Aligner won't run" % (self._mapper.batch_size, len(reads) * 2))
for idx, fragment in enumerate(reads):
self._ctx.set_input_key(idx * 100)
self._ctx.set_input_value('\t'.join(fragment))
self._mapper.map(self._ctx)
produced_sam = sorted(self._ctx.sam_lines)
self.assertEquals(len(expected_output), len(produced_sam))
self.assertEquals(expected_output, produced_sam)
self.assertEquals(len(reads) * 2, self._ctx.counters["SEQAL:EMITTED SAM RECORDS"])
def test_simple_map(self):
# get input data and expected output. We keep exactly one map batch of reads
# (as per self._map_batch_size)
reads = tseal_utils.get_mini_ref_seqs()[0:(self._map_batch_size / 2)]
expected_output = sorted(
tseal_utils.rapi_mini_ref_seqs_sam_no_header().split(
'\n')[0:2 * len(reads)])
self._log.info("loaded %s fragments and %s lines of expected output",
len(reads), len(expected_output))
if len(reads) * 2 < self._mapper.batch_size:
self.fail(
"batch size for test (%s) is set larger than the number of available "
"reads (%s). Aligner won't run" %
(self._mapper.batch_size, len(reads) * 2))
for idx, fragment in enumerate(reads):
self._ctx.set_input_key(idx * 100)
self._ctx.set_input_value('\t'.join(fragment))
self._mapper.map(self._ctx)
produced_sam = sorted(self._ctx.sam_lines)
self.assertEquals(len(expected_output), len(produced_sam))
self.assertEquals(expected_output, produced_sam)
self.assertEquals(
len(reads) * 2, self._ctx.counters["SEQAL:EMITTED SAM RECORDS"])
