def test_duplicate_input_ids_across_aligned_and_unaligned(self):
input_fp = self.get_data_path('aligned-duplicate-ids-2.fasta')
alignment = DNAFASTAFormat(input_fp, mode='r')
_, sequences, _ = self._prepare_sequence_data()
with self.assertRaisesRegex(ValueError, 'aligned and unaligned.*seq1'):
with redirected_stdio(stderr=os.devnull):
mafft_add(alignment, sequences)
def test_duplicate_input_ids_in_unaligned(self):
input_fp = self.get_data_path('unaligned-duplicate-ids.fasta')
sequences = DNAFASTAFormat(input_fp, mode='r')
alignment, _, _ = self._prepare_sequence_data()
with self.assertRaisesRegex(ValueError, 'the unaligned.*id1'):
with redirected_stdio(stderr=os.devnull):
mafft_add(alignment, sequences)
def test_mafft_add_flags(self):
alignment, sequences, exp = self._prepare_sequence_data()
with patch('q2_alignment._mafft.run_command') as patched_run_cmd:
with patch('q2_alignment._mafft.skbio.TabularMSA.read',
return_value=exp):
_ = mafft_add(alignment, sequences)
patched_run_cmd.assert_called_with([
"mafft", "--preservecase", "--inputorder", "--thread", "1",
"--add", ANY, ANY
], ANY)
_ = mafft_add(alignment, sequences, addfragments=True)
patched_run_cmd.assert_called_with([
"mafft", "--preservecase", "--inputorder", "--thread", "1",
"--addfragments", ANY, ANY
], ANY)
def test_mafft_add_fragments(self):
alignment, sequences, exp = self._prepare_sequence_data()
with redirected_stdio(stderr=os.devnull):
result = mafft_add(alignment, sequences, addfragments=True)
obs = skbio.io.read(str(result),
into=skbio.TabularMSA,
constructor=skbio.DNA)
self.assertEqual(obs, exp)
def test_long_ids_are_not_truncated_aligned(self):
input_fp = self.get_data_path('aligned-long-ids.fasta')
alignment = DNAFASTAFormat(input_fp, mode='r')
_, sequences, _ = self._prepare_sequence_data()
with redirected_stdio(stderr=os.devnull):
result = mafft_add(alignment, sequences)
with open(str(result), 'r') as fh:
obs = fh.read()
self.assertIn('a' * 250, obs)
self.assertIn('b' * 250, obs)
self.assertIn('seq1', obs)
self.assertIn('seq2', obs)