def test_fasta_to_generator_valid_files(self):
test_cases = (self.empty, self.single, self.multi,
self.odd_labels_different_type,
self.sequence_collection_different_type,
self.lowercase_seqs)
# Strategy:
# for each fasta file, read it without its corresponding qual file,
# and ensure observed vs. expected match, ignoring quality scores in
# expected. next, parse the current fasta file with each
# corresponding quality file and ensure that observed vs. expected
# match, this time taking quality scores into account. this
# sufficiently exercises parsing a standalone fasta file and paired
# fasta/qual files
for exp, kwargs, fasta_fps, qual_fps in test_cases:
for fasta_fp in fasta_fps:
obs = list(_fasta_to_generator(fasta_fp, **kwargs))
self.assertEqual(len(obs), len(exp))
for o, e in zip(obs, exp):
e = e.copy()
del e.positional_metadata['quality']
self.assertEqual(o, e)
for qual_fp in qual_fps:
obs = list(_fasta_to_generator(fasta_fp, qual=qual_fp,
**kwargs))
self.assertEqual(len(obs), len(exp))
for o, e in zip(obs, exp):
self.assertEqual(o, e)
def test_roundtrip_generators(self):
# test that fasta and qual files can be streamed into memory and back
# out to disk using generator reader and writer
fps = list(map(lambda e: list(map(get_data_path, e)),
[('empty', 'empty'),
('fasta_multi_seq_roundtrip',
'qual_multi_seq_roundtrip')]))
for fasta_fp, qual_fp in fps:
with io.open(fasta_fp) as fh:
exp_fasta = fh.read()
with io.open(qual_fp) as fh:
exp_qual = fh.read()
fasta_fh = io.StringIO()
qual_fh = io.StringIO()
_generator_to_fasta(_fasta_to_generator(fasta_fp, qual=qual_fp),
fasta_fh, qual=qual_fh)
obs_fasta = fasta_fh.getvalue()
obs_qual = qual_fh.getvalue()
fasta_fh.close()
qual_fh.close()
self.assertEqual(obs_fasta, exp_fasta)
self.assertEqual(obs_qual, exp_qual)
def _construct_seq(fh, constructor=DNA, seq_num=1):
lines = []
for i, (data_type, seq_id, l) in enumerate(_yield_record(fh), 1):
if data_type == 'data' and seq_num == i:
lines = l
seq = _get_nth_sequence(_fasta_to_generator(fh, constructor=constructor),
seq_num=seq_num)
seq.interval_metadata = _parse_record(lines, len(seq))
return seq
def _construct_seq(fh, constructor=DNA, seq_num=1):
lines = []
for i, (data_type, seq_id, l) in enumerate(_yield_record(fh), 1):
if data_type == 'data' and seq_num == i:
lines = l
seq = _get_nth_sequence(_fasta_to_generator(fh, constructor=constructor),
seq_num=seq_num)
seq.interval_metadata = _parse_record(lines, len(seq))
return seq
def test_fasta_to_generator_invalid_files(self):
for fp, kwargs, error_type, error_msg_regex in self.invalid_fps:
with six.assertRaisesRegex(self, error_type, error_msg_regex):
list(_fasta_to_generator(fp, **kwargs))
