def testMaxSequenceLength(self):
a = Alignment()
a['1'] = 'A--CG--T'
a['2'] = 'AC----GT'
a['3'] = 'A-C-G-T-'
a['4'] = 'ACGT---T'
self.assertEqual(a.max_sequence_length(), 5)
def testMaxSequenceLength(self):
a = Alignment()
a['1'] = 'A--CG--T'
a['2'] = 'AC----GT'
a['3'] = 'A-C-G-T-'
a['4'] = 'ACGT---T'
self.assertEqual(a.max_sequence_length(), 5)
def setUp(self):
self.ts = TempFS()
self.ts.create_top_level_temp(prefix='treeEstimatorTest', parent=os.curdir)
self.filename = data_source_path('mafft.anolis.fasta')
self.alignment = Alignment()
self.alignment.read_filepath(data_source_path('mafft.anolis.fasta'),
'FASTA')
def testConcatenateAlignments(self):
filename1 = data_source_path('small.fasta')
filename2 = data_source_path('small.fasta')
a = Alignment()
b = Alignment()
a.datatype = "DNA"
b.datatype = "DNA"
a.read_filepath(filename1, 'FASTA')
b.read_filepath(filename2, 'FASTA')
def _impl_test_aligner(self, name, fn):
filename = data_source_path(fn)
alignment = Alignment()
alignment.read_filepath(filename, 'FASTA')
aln = self.get_aligner('%s' % name)
if aln is None:
_LOG.warn("test%s skipped" % name)
return
a = aln.run(alignment,
tmp_dir_par=self.ts.top_level_temp,
delete_temps=True)
reference_fn = data_source_path('%s.%s' % (name, fn))
reference_aln = Alignment()
reference_aln.read_filepath(reference_fn, 'FASTA')
_LOG.debug('Checking results from %s against %s' % (name, reference_fn))
if reference_aln != a:
i = 1
while True:
nrfn = reference_fn + '.' + str(i)
if os.path.exists(nrfn):
reference_aln = Alignment()
reference_aln.read_filepath(nrfn, 'FASTA')
_LOG.debug('Checking results from %s against %s' % (name, nrfn))
if reference_aln == a:
self.assertEquals(reference_aln, a)
return True
i += 1
else:
self.assertEquals(reference_aln, a)
def testConcatenateAlignments(self):
filename1 = data_source_path('small.fasta')
filename2 = data_source_path('small.fasta')
a = Alignment()
b = Alignment()
a.datatype = "DNA"
b.datatype = "DNA"
a.read_filepath(filename1, 'FASTA')
b.read_filepath(filename2, 'FASTA')
def _impl_test_merger(self, name):
filename = data_source_path('merger1.fasta')
alignment1 = Alignment()
alignment1.read_filepath(filename, 'FASTA')
filename = data_source_path('merger2.fasta')
alignment2 = Alignment()
alignment2.read_filepath(filename, 'FASTA')
aln = self.get_merger('%s merger' % name)
if aln is None:
_LOG.warn("test%s skipped" % name)
return
a = aln.run(alignment1,
alignment2,
tmp_dir_par=self.ts.top_level_temp,
delete_temps=True)
reference_fn = data_source_path('merger_result.fasta')
reference_aln = Alignment()
reference_aln.read_filepath(reference_fn, 'FASTA')
self.assertEquals(reference_aln, a)
def testAlignment(self):
filename = data_source_path('small.fasta')
alignment = Alignment()
alignment.read_filepath(filename, 'FASTA')
num_taxa = alignment.get_num_taxa()
self.assertEqual(num_taxa, 32)
alignment.write_filepath(filename + '.phy', 'PHYLIP')
alignment.write_unaligned_fasta(filename + '.raw')
alignment.sub_alignment(list(
alignment.keys())[0:2]).write_unaligned_fasta(filename +
'.partial.raw')
def _impl_test_aligner(self, name, fn):
filename = data_source_path(fn)
alignment = Alignment()
alignment.read_filepath(filename, 'FASTA')
aln = self.get_aligner('%s' % name)
if aln is None:
_LOG.warn("test%s skipped" % name)
return
a = aln.run(alignment,
tmp_dir_par=self.ts.top_level_temp,
delete_temps=True)
reference_fn = data_source_path('%s.%s' % (name, fn))
reference_aln = Alignment()
reference_aln.read_filepath(reference_fn, 'FASTA')
_LOG.debug('Checking results from %s against %s' %
(name, reference_fn))
if reference_aln != a:
i = 1
while True:
nrfn = reference_fn + '.' + str(i)
if os.path.exists(nrfn):
reference_aln = Alignment()
reference_aln.read_filepath(nrfn, 'FASTA')
_LOG.debug('Checking results from %s against %s' %
(name, nrfn))
if reference_aln == a:
self.assertEquals(reference_aln, a)
return True
i += 1
else:
self.assertEquals(reference_aln, a)
def testAlignment(self):
filename = data_source_path('small.fasta')
alignment = Alignment()
alignment.read_filepath(filename, 'FASTA')
num_taxa = alignment.get_num_taxa()
self.assertEqual(num_taxa, 32)
alignment.write_filepath(filename+'.phy', 'PHYLIP')
alignment.write_unaligned_fasta(filename+'.raw')
alignment.sub_alignment( alignment.keys()[0:2] ).write_unaligned_fasta(filename+'.partial.raw')
class TreeEstimatorTest(unittest.TestCase):
def setUp(self):
self.ts = TempFS()
self.ts.create_top_level_temp(prefix='treeEstimatorTest', parent=os.curdir)
self.filename = data_source_path('mafft.anolis.fasta')
self.alignment = Alignment()
self.alignment.read_filepath(data_source_path('mafft.anolis.fasta'),
'FASTA')
def tearDown(self):
dir_list = self.ts.get_remaining_directories()
for dir in dir_list:
try:
self.ts.remove_dir(dir)
except ValueError:
pass
def get_tree_estimator(self, name):
try:
return config.create_tree_estimator(name=name, temp_fs=self.ts)
except RuntimeError:
log_exception(_LOG)
_LOG.warn("""Could not create an aligner of type %s !
This could indicate a bug in create_tree_estimator_using_config() or could mean that
your installation is not configured to run this tool.
""" % name)
return None
def _impl_test_tree_estimator(self, name, datatype, partitions, **kwargs):
num_cpus = kwargs.get('num_cpus', None)
filename = data_source_path('anolis.fasta')
md = MultiLocusDataset()
md.read_files(seq_filename_list=[filename],
datatype=datatype)
md.relabel_for_pasta()
# alignment = Alignment()
# alignment.read_filepath(filename, 'FASTA')
te = self.get_tree_estimator(name)
if te is None:
_LOG.warn("test%s skipped" % name)
return
# alignment.datatype = datatype
if num_cpus:
a = te.run(alignment=md,
partitions=partitions,
tmp_dir_par=self.ts.top_level_temp,
delete_temps=True,
num_cpus=num_cpus)
else:
a = te.run(alignment=md,
partitions=partitions,
tmp_dir_par=self.ts.top_level_temp,
delete_temps=True)
def testRaxml(self):
# if is_test_enabled(TestLevel.SLOW, _LOG):
self._impl_test_tree_estimator('raxml', datatype="DNA",
partitions=[("DNA", 1, 1456)])
def testFastTree(self):
config.fasttree.add_option('options', StringUserSetting(
name='options',
default='',
short_name=None,
help='Options to be passed to FastTree.',
subcategory=None))
self._impl_test_tree_estimator('fasttree', datatype="DNA",
partitions=[("DNA", 1, 1456)])
def testFastTreeMP(self):
config.fasttree.add_option('options', StringUserSetting(
name='options',
default='',
short_name=None,
help='Options to be passed to FastTree.',
subcategory=None))
self._impl_test_tree_estimator('fasttree', datatype="DNA",
partitions=[("DNA", 1, 1456)], num_cpus=2)
def _impl_test_merger(self, name):
filename = data_source_path('merger1.fasta')
alignment1 = Alignment()
alignment1.read_filepath(filename, 'FASTA')
filename = data_source_path('merger2.fasta')
alignment2 = Alignment()
alignment2.read_filepath(filename, 'FASTA')
aln = self.get_merger('%s merger' % name)
if aln is None:
_LOG.warn("test%s skipped" % name)
return
a = aln.run(alignment1,
alignment2,
tmp_dir_par=self.ts.top_level_temp,
delete_temps=True)
reference_fn = data_source_path('merger_result.fasta')
reference_aln = Alignment()
reference_aln.read_filepath(reference_fn, 'FASTA')
self.assertEquals(reference_aln, a)
