def test_empty_table(self):
table = biom.Table(np.array([[]]), [], [])
core_features(self.output_dir, table)
self.assertBasicVizValidity(self.output_dir, exp_sample_count=0,
exp_feature_count=0, exp_no_core=True)
self.assertFalse(glob.glob(os.path.join(self.output_dir, '*.tsv')))
def test_defaults(self):
core_features(self.output_dir, self.table)
self.assertBasicVizValidity(self.output_dir)
core_50_fp = os.path.join(self.output_dir, 'core-features-0.500.tsv')
self.assertCoreFeaturesPresent(core_50_fp, ['O1', 'O2'], [])
core_100_fp = os.path.join(self.output_dir, 'core-features-1.000.tsv')
self.assertCoreFeaturesPresent(core_100_fp, ['O2'], ['O1'])
def test_equal_min_max_fractions(self):
core_features(self.output_dir, self.table, min_fraction=0.55,
max_fraction=0.55)
self.assertBasicVizValidity(self.output_dir, exp_no_plot=True)
core_55_fp = os.path.join(self.output_dir, 'core-features-0.550.tsv')
tsv_files = glob.glob(os.path.join(self.output_dir, '*.tsv'))
self.assertEqual(tsv_files, [core_55_fp])
self.assertCoreFeaturesPresent(core_55_fp, ['O1', 'O2'], [])
def test_minimum_number_of_steps(self):
core_features(self.output_dir, self.table, min_fraction=0.55,
max_fraction=0.95, steps=2)
self.assertBasicVizValidity(self.output_dir)
core_55_fp = os.path.join(self.output_dir, 'core-features-0.550.tsv')
core_95_fp = os.path.join(self.output_dir, 'core-features-0.950.tsv')
tsv_files = sorted(glob.glob(os.path.join(self.output_dir, '*.tsv')))
self.assertEqual(tsv_files, [core_55_fp, core_95_fp])
self.assertCoreFeaturesPresent(core_55_fp, ['O1', 'O2'], [])
self.assertCoreFeaturesPresent(core_95_fp, ['O2'], ['O1'])
def test_no_core_features(self):
table = biom.Table(np.array([[0, 11, 11], [11, 11, 0]]), ['O1', 'O2'],
['S1', 'S2', 'S3'])
core_features(self.output_dir, table)
self.assertBasicVizValidity(self.output_dir, exp_no_core=True)
core_50_fp = os.path.join(self.output_dir, 'core-features-0.500.tsv')
self.assertCoreFeaturesPresent(core_50_fp, ['O1', 'O2'], [])
# No core features exist at fraction=1.0
core_100_fp = os.path.join(self.output_dir, 'core-features-1.000.tsv')
self.assertFalse(os.path.exists(core_100_fp))
def test_fraction_range(self):
core_features(self.output_dir, self.table, min_fraction=0.55,
max_fraction=0.95, steps=9)
self.assertBasicVizValidity(self.output_dir)
core_50_fp = os.path.join(self.output_dir, 'core-features-0.500.tsv')
self.assertFalse(os.path.exists(core_50_fp))
core_55_fp = os.path.join(self.output_dir, 'core-features-0.550.tsv')
self.assertCoreFeaturesPresent(core_55_fp, ['O1', 'O2'], [])
core_95_fp = os.path.join(self.output_dir, 'core-features-0.950.tsv')
self.assertCoreFeaturesPresent(core_95_fp, ['O2'], ['O1'])
core_100_fp = os.path.join(self.output_dir, 'core-features-1.000.tsv')
self.assertFalse(os.path.exists(core_100_fp))
def test_tiny_steps_precision(self):
# Bug in a previous iteration of the code used a fixed number of digits
# in the "feature list" TSV filenames, causing files to be overwritten
# with the wrong data if a small enough step size was used.
core_features(self.output_dir, self.table, min_fraction=0.1,
max_fraction=0.11, steps=101)
self.assertBasicVizValidity(self.output_dir)
fp = os.path.join(self.output_dir, 'core-features-0.1000.tsv')
self.assertTrue(os.path.exists(fp))
fp = os.path.join(self.output_dir, 'core-features-0.1001.tsv')
self.assertTrue(os.path.exists(fp))
fp = os.path.join(self.output_dir, 'core-features-0.1099.tsv')
self.assertTrue(os.path.exists(fp))
fp = os.path.join(self.output_dir, 'core-features-0.1100.tsv')
self.assertTrue(os.path.exists(fp))
def test_invalid_parameters(self):
with self.assertRaisesRegex(ValueError, 'fraction'):
core_features(self.output_dir,
self.table,
min_fraction=0.75,
max_fraction=0.5)
