def test_rev_comp_gappy_muscle(self):
if not functions.which('muscle'):
_LOG.warning('muscle not found... skipping tests.')
return
self.rc_seqs = dataio.get_buffered_seq_iter([self.gappy_path])
d, e = seqsum.summarize_distances(self.rc_seqs,
sample_size=5,
per_site=False,
aligned=False,
ignore_gaps=True,
do_full_alignment=False,
aligner_tools=['muscle'],
full_aligner_tools=None)
self.assertTrue(len(e) < 1)
self.assertEqual(len(d), 31)
def align(self, seq_iter):
with TemporaryFilePath() as tmp_path:
with OpenFile(tmp_path, 'w') as tmp:
for seq in seq_iter:
tmp.write(seq.format('fasta'))
self.kwargs['input'] = tmp_path
mafft_command = MafftCommandline(self.exe, **self.kwargs)
self.cmd = str(mafft_command)
_LOG.debug('{0}: Executing command {1!r}'.format(
self.name, self.cmd))
stdout, stderr = mafft_command()
if self.out_path:
self.out_path = functions.get_new_path(self.out_path)
with OpenFile(self.out_path, 'w') as out:
out.write(stdout)
return dataio.get_buffered_seq_iter([StringIO(stdout)], format='fasta')
def align(self, seq_iter):
with TemporaryFilePath() as tmp_path:
with OpenFile(tmp_path, 'w') as tmp:
for seq in seq_iter:
tmp.write(seq.format('fasta'))
self.kwargs['input'] = tmp_path
mafft_command = MafftCommandline(self.exe, **self.kwargs)
self.cmd = str(mafft_command)
_LOG.debug('{0}: Executing command {1!r}'.format(self.name,
self.cmd))
stdout, stderr = mafft_command()
if self.out_path:
self.out_path = functions.get_new_path(self.out_path)
with OpenFile(self.out_path, 'w') as out:
out.write(stdout)
return dataio.get_buffered_seq_iter([StringIO(stdout)], format='fasta')
def test_rev_comp_error_mafft_full(self):
if not functions.which('mafft'):
_LOG.warning('mafft not found... skipping tests.')
return
self.rc_seqs = dataio.get_buffered_seq_iter([self.rc_path])
d, e = seqsum.summarize_distances(self.rc_seqs,
sample_size=0,
per_site=False,
aligned=False,
ignore_gaps=False,
do_full_alignment=True,
aligner_tools=['mafft'],
full_aligner_tools=['mafft'])
self.assertEqual(len(e), 11)
for rce in e:
self.assertTrue('Homo_sapiens' in rce)
self.assertEqual(len(d), 12)
def align(self, seq_iter):
with TemporaryFilePath() as in_path:
with OpenFile(in_path, 'w') as tmp:
for seq in seq_iter:
tmp.write(seq.format('fasta'))
self.kwargs['input'] = in_path
with TemporaryFilePath() as tmp_out_path:
self.kwargs['out'] = tmp_out_path
muscle_command = MuscleCommandline(self.exe, **self.kwargs)
self.cmd = str(muscle_command)
_LOG.debug('{0}: Executing command {1!r}'.format(
self.name, self.cmd))
stdout, stderr = muscle_command()
results = dataio.get_buffered_seq_iter([tmp_out_path],
format='fasta')
if self.out_path:
self.out_path = functions.get_new_path(self.out_path)
shutil.move(tmp_out_path, self.out_path)
return results
def align(self, seq_iter):
with TemporaryFilePath() as in_path:
with OpenFile(in_path, 'w') as tmp:
for seq in seq_iter:
tmp.write(seq.format('fasta'))
self.kwargs['input'] = in_path
with TemporaryFilePath() as tmp_out_path:
self.kwargs['out'] = tmp_out_path
muscle_command = MuscleCommandline(self.exe, **self.kwargs)
self.cmd = str(muscle_command)
_LOG.debug('{0}: Executing command {1!r}'.format(self.name,
self.cmd))
stdout, stderr = muscle_command()
results = dataio.get_buffered_seq_iter([tmp_out_path],
format='fasta')
if self.out_path:
self.out_path = functions.get_new_path(self.out_path)
shutil.move(tmp_out_path, self.out_path)
return results