def test_duplicate_input_ids(self):
input_fp = self.get_data_path('unaligned-duplicate-ids.fasta')
input_sequences = DNAFASTAFormat(input_fp, mode='r')
with self.assertRaisesRegex(ValueError, 'the unaligned.*id1'):
with redirected_stdio(stderr=os.devnull):
mafft(input_sequences)
def test_mafft_parttree_exception(self):
input_fp = os.path.join(self.temp_dir.name, 'million.fasta')
with open(input_fp, "w") as f:
for i in range(0, 1000002):
f.write('>%d\nAAGCAAGC\n' % i)
input_sequences = DNAFASTAFormat(input_fp, mode='r')
with self.assertRaisesRegex(ValueError, '1 million'):
with redirected_stdio(stderr=os.devnull):
mafft(input_sequences)
def test_multithreaded_mafft(self):
input_sequences, exp = self._prepare_sequence_data()
with redirected_stdio(stderr=os.devnull):
result = mafft(input_sequences, n_threads='auto')
obs = skbio.io.read(str(result), into=skbio.TabularMSA,
constructor=skbio.DNA)
self.assertEqual(obs, exp)
def test_mafft(self):
input_fp = self.get_data_path('unaligned-dna-sequences-1.fasta')
input_sequences = DNAFASTAFormat(input_fp, mode='r')
exp = skbio.TabularMSA(
[skbio.DNA('AGGGGGG', metadata={'id': 'seq1', 'description': ''}),
skbio.DNA('-GGGGGG', metadata={'id': 'seq2', 'description': ''})]
)
with redirected_stdio(stderr=os.devnull):
result = mafft(input_sequences)
obs = skbio.io.read(str(result), into=skbio.TabularMSA,
constructor=skbio.DNA)
self.assertEqual(obs, exp)
def test_long_ids_are_not_truncated(self):
input_fp = self.get_data_path('unaligned-long-ids.fasta')
input_sequences = DNAFASTAFormat(input_fp, mode='r')
with redirected_stdio(stderr=os.devnull):
result = mafft(input_sequences)
with open(str(result), 'r') as fh:
obs = fh.read()
self.assertIn('a' * 250, obs)
self.assertIn('b' * 250, obs)
self.assertIn('c' * 250, obs)
def test_mafft(self):
input_fp = self.get_data_path('unaligned-dna-sequences-1.fasta')
input_sequences = DNAFASTAFormat(input_fp, mode='r')
exp = skbio.TabularMSA([
skbio.DNA('AGGGGGG', metadata={
'id': 'seq1',
'description': ''
}),
skbio.DNA('-GGGGGG', metadata={
'id': 'seq2',
'description': ''
})
])
with redirected_stdio(stderr=os.devnull):
result = mafft(input_sequences)
obs = skbio.io.read(str(result),
into=skbio.TabularMSA,
constructor=skbio.DNA)
self.assertEqual(obs, exp)
