def test_clustal_outfile(self):
out = self.seq.replace('sequence.fa', 'msa.out.clustal.fa')
aln = msa(CLUSTAL, self.seq, outfile=out)
self.assertTrue(os.path.isfile(aln))
self.assertEqual(out, aln)
self.rm = aln
def test_clustal_default(self):
aln = msa(CLUSTAL, self.seq)
self.assertTrue(os.path.isfile(aln))
self.rm = aln
def test_mafft_outfile(self):
out = self.seq.replace('sequence.fa', 'msa.out.mafft.fa')
aln = msa(MAFFT, self.seq, outfile=out)
self.assertTrue(os.path.isfile(aln))
self.assertEqual(out, aln)
self.rm = aln
def test_mafft_default(self):
aln = msa(MAFFT, self.seq)
self.assertTrue(os.path.isfile(aln))
self.rm = aln
def test_muscle_default(self):
aln = msa(MUSCLE, self.seq)
self.assertTrue(os.path.isfile(aln))
self.rm = aln
def _sequencing(sequence, tree, aligner, ancestor, wd, asr_model, verbose):
"""
Identify the type of the sequence file.
:param sequence: str, path to a sequence data file.
:param tree: str, path to a NEWICK tree file.
:return: tuple, sequence, alignment, ancestor, and simulation data file.
"""
if tree:
utilities.Tree(tree, leave=True)
AA, lengths, aa = set(AMINO_ACIDS), [], []
with open(sequence) as handle:
line = handle.readline().strip()
if line.startswith('>'):
handle.seek(0)
records = SeqIO.parse(handle, 'fasta')
for record in records:
lengths.append(len(record.seq))
aa.append(set(record.seq).issubset(AA))
else:
error('NEWICK format tree was provided, but the sequence file '
'was not in the FASTA format.')
sys.exit(1)
if len(set(lengths)) == 1:
alignment = sequence
if all(aa):
trimmed = alignment
else:
trimmed = ''.join(
[utilities.basename(alignment), '.trimmed.fasta'])
if os.path.isfile(trimmed):
info('Using pre-existed trimmed alignment file.')
else:
_, trimmed = utilities.trim(alignment, outfile=trimmed)
else:
if aligner:
aler, _ = msa._guess(aligner)
outfile = ''.join(
[utilities.basename(sequence), '.{}.fasta'.format(aler)])
if os.path.isfile(outfile):
info('Using pre-existed alignment file')
alignment = outfile
trimmed = ''.join(
[utilities.basename(alignment), '.trimmed.fasta'])
if os.path.isfile(trimmed):
info('Using pre-existed trimmed alignment file.')
else:
_, trimmed = utilities.trim(alignment, outfile=trimmed)
else:
trimmed = msa.msa(aligner,
sequence,
verbose=verbose,
outfile=outfile,
trimming=True)
else:
error('FASTA format sequence file was provided, but no '
'alignment program was provided.')
sys.exit(1)
if trimmed:
if ancestor:
if trimmed.endswith('.trimmed.fasta'):
name = trimmed.replace('.trimmed.fasta', '')
else:
name = trimmed
aser, _ = asr._guess(ancestor)
outfile = '{}.{}.tsv'.format(utilities.basename(name), aser)
if os.path.isfile(outfile):
info('Using pre-existed ancestral states sequence file.')
sequence = outfile
else:
sequence = asr.asr(ancestor,
trimmed,
tree,
asr_model,
verbose=verbose,
outfile=outfile)
else:
error('No ancestral reconstruction program was provided.')
sys.exit(1)
else:
sys.exit(1)
tree, rate, records, aps, size = _load(sequence)
return tree, rate, records, aps, size, sequence