def test_get_labels(self):
a = Connectivity(depth=2)
X = a.fit_transform([METHANE])
labels = a.get_labels()
self.assertEqual(X.shape[1], len(labels))
expected = ('C-H', )
self.assertEqual(labels, expected)
def test_fit_atom_coordination(self):
a = Connectivity(depth=1, use_coordination=True)
a.fit(ALL_DATA)
self.assertEqual(
a._base_chains,
set([('C1', ), ('N3', ), ('N2', ), ('O2', ), ('N1', ), ('O1', ),
('C4', ), ('H0', ), ('H1', ), ('O0', ), ('C3', ), ('C2', )]))
def test_get_labels_bond_order(self):
a = Connectivity(depth=3, use_bond_order=True)
X = a.fit_transform([METHANE])
labels = a.get_labels()
self.assertEqual(X.shape[1], len(labels))
expected = ('H-C-1_C-H-1', )
self.assertEqual(labels, expected)
def test_unknown(self):
a = Connectivity(add_unknown=True)
expected_results = numpy.array([[1, 4, 0],
[2, 3, 4],
[25, 15, 9]])
a.fit([METHANE])
self.assertTrue((a.transform(ALL_DATA) == expected_results).all())
def test_get_labels_coordination(self):
a = Connectivity(depth=1, use_coordination=True)
X = a.fit_transform([METHANE])
labels = a.get_labels()
self.assertEqual(X.shape[1], len(labels))
expected = ('C4', 'H1')
self.assertEqual(labels, expected)
def test_fit_bond(self):
a = Connectivity(depth=2)
a.fit(ALL_DATA)
self.assertEqual(
a._base_chains,
set([('H', 'O'), ('C', 'H'), ('H', 'N'), ('C', 'C'), ('H', 'H'),
('O', 'O'), ('C', 'N'), ('C', 'O')]))
def test_fit_atom_bond(self):
# This should be the exact same thing as doing it with
# use_bond_order=False
a = Connectivity(depth=1, use_bond_order=True)
a.fit(ALL_DATA)
self.assertEqual(a._base_chains,
set([('N', ), ('C', ), ('O', ), ('H', )]))
def test_get_labels_unknown(self):
a = Connectivity(depth=2, add_unknown=True)
X = a.fit_transform([METHANE])
labels = a.get_labels()
self.assertEqual(X.shape[1], len(labels))
expected = ('C-H', UNKNOWN)
self.assertEqual(labels, expected)
def test_fit_angle(self):
a = Connectivity(depth=3)
a.fit(ALL_DATA)
self.assertEqual(a._base_groups,
(('C', 'C', 'C'), ('C', 'C', 'H'), ('C', 'C', 'N'),
('C', 'C', 'O'), ('C', 'N', 'C'), ('C', 'N', 'H'),
('C', 'O', 'C'), ('C', 'O', 'H'), ('H', 'C', 'H'),
('H', 'C', 'N'), ('H', 'C', 'O'), ('H', 'N', 'H'),
('N', 'C', 'N'), ('N', 'C', 'O')))
def test_fit_bond_bond(self):
a = Connectivity(depth=2, use_bond_order=True)
a.fit(ALL_DATA)
self.assertEqual(
a._base_groups,
((('C', 'C', '1'), ), (('C', 'C', '2'), ), (('C', 'C', '3'), ),
(('C', 'C', 'Ar'), ), (('C', 'H', '1'), ), (('C', 'N', '2'), ),
(('C', 'N', '3'), ), (('C', 'N', 'Ar'), ), (('C', 'O', '1'), ),
(('C', 'O', 'Ar'), ), (('H', 'H', '1'), ), (('H', 'N', '1'), ),
(('H', 'O', '1'), ), (('O', 'O', '1'), )))
def test_tfidf(self):
a = Connectivity(do_tfidf=True)
expected = numpy.array([[0., 0., 0., 0.], [0., 0., 0., 1.62186043],
[0., 0., 5.49306144, 1.62186043]])
a.fit(ALL_DATA)
try:
m = a.transform(ALL_DATA)
numpy.testing.assert_array_almost_equal(m, expected)
except AssertionError as e:
self.fail(e)
def test_fit_dihedral(self):
# This is to test the double order flipping (CCCH vs HCCC)
a = Connectivity(depth=4)
a.fit(ALL_DATA)
self.assertEqual(
a._base_groups,
(('C', 'C', 'C', 'C'), ('C', 'C', 'C', 'H'), ('C', 'C', 'C', 'N'),
('C', 'C', 'C', 'O'), ('C', 'C', 'N', 'C'), ('C', 'C', 'N', 'H'),
('C', 'C', 'O', 'C'), ('C', 'C', 'O', 'H'), ('H', 'C', 'C', 'H'),
('H', 'C', 'C', 'N'), ('H', 'C', 'C', 'O'), ('H', 'C', 'N', 'C'),
('H', 'C', 'O', 'C'), ('N', 'C', 'C', 'O'), ('N', 'C', 'N', 'C'),
('N', 'C', 'N', 'H'), ('N', 'C', 'O', 'H'), ('O', 'C', 'N', 'C')))
def test_fit(self):
trans = Connectivity(input_type="filename")
a = FragmentMap(transformer=trans)
a.fit([ALL])
expected = {'mid': [2, 3, 4], 'methane': [1, 4, 0]}
simplified = {x: y.tolist() for x, y in a._x_fragments.items()}
self.assertEqual(expected, simplified)
def test_callable_filename_to_label(self):
trans = Connectivity(input_type="filename")
a = FragmentMap(transformer=trans,
filename_to_label=lambda x: x[-5:],
label_to_filename=(DATA_PATH, ))
func = a._get_filename_to_label()
self.assertEqual(func(METHANE_PATH), METHANE_PATH[-5:])
def test_get_labels(self):
trans = Connectivity(input_type="filename")
a = FragmentMap(transformer=trans)
res = a.fit_transform([ALL])
labels = a.get_labels()
self.assertEqual(res.shape[2], len(labels))
expected = ('C', 'H', 'O')
self.assertEqual(labels, expected)
def test_get_labels_no_labels(self):
trans = Pipeline([('Con', Connectivity(input_type="filename"))])
a = FragmentMap(transformer=trans)
res = a.fit_transform([ALL])
labels = a.get_labels()
self.assertEqual(res.shape[2], len(labels))
expected = ('0', '1', '2')
self.assertEqual(labels, expected)
def test_fit_transform(self):
trans = Connectivity(input_type="filename")
a = FragmentMap(transformer=trans)
res = a.fit_transform([ALL])
expected = numpy.array([[[1, 4, 0], [2, 3, 4]]])
try:
numpy.testing.assert_array_almost_equal(expected, res)
except AssertionError as e:
self.fail(e)
def test_callable_label_to_filename(self):
trans = Connectivity(input_type="filename")
def func(x):
return os.path.join(DATA_PATH, x)
a = FragmentMap(transformer=trans, label_to_filename=func)
self.assertEqual(a._get_label_to_filename()('test'),
os.path.join(DATA_PATH, 'test'))
def test_label_to_filename(self):
trans = Connectivity(input_type="filename")
a = FragmentMap(transformer=trans, label_to_filename=(DATA_PATH, ))
# The paths are sorted when searching
path = METHANE_PATH.replace('.out', '.cry')
self.assertEqual(a._get_label_to_filename()('methane'), path)
def test_label_to_filename_not_found(self):
trans = Connectivity(input_type="filename")
a = FragmentMap(transformer=trans, label_to_filename=(DATA_PATH, ))
with self.assertRaises(ValueError):
a._get_label_to_filename()('not real')
def test_fit_atom_separated(self):
a = Connectivity(depth=1)
a.fit([METHANE2])
self.assertEqual(a._base_chains, set([('C', ), ('H', )]))
self.assertTrue((a.transform([METHANE2]) == numpy.array([[1,
4]])).all())
def test_fit_atom(self):
a = Connectivity(depth=1)
a.fit(ALL_DATA)
self.assertEqual(a._base_chains,
set([('N', ), ('C', ), ('O', ), ('H', )]))
def test_fit_transform(self):
a = Connectivity()
self.assertTrue((a.fit_transform(ALL_DATA) == ALL_ATOM).all())
def test_large_to_small_transform(self):
a = Connectivity()
a.fit([BIG])
self.assertTrue((a.transform(ALL_DATA) == ALL_ATOM).all())
def test_transform_before_fit(self):
a = Connectivity()
with self.assertRaises(ValueError):
a.transform(ALL_DATA)
def test_small_to_large_transform(self):
a = Connectivity()
a.fit([METHANE])
self.assertTrue((a.transform(ALL_DATA) == ALL_ATOM[:, :2]).all())
def test_fit_atom(self):
a = Connectivity(depth=1)
a.fit(ALL_DATA)
self.assertEqual(a._base_groups, (('C',), ('H',), ('N',), ('O',)))
def test_invalid_label_to_filename(self):
trans = Connectivity(input_type="filename")
a = FragmentMap(transformer=trans, label_to_filename=lambda x: 1)
with self.assertRaises(KeyError):
a.fit_transform(ALL)
def test_bad_input_type(self):
trans = Connectivity(input_type="filename")
a = FragmentMap(input_type='bad', transformer=trans)
with self.assertRaises(ValueError):
a.fit([ALL])
def test_label_input_type(self):
trans = Connectivity(input_type="filename")
a = FragmentMap(input_type='label',
transformer=trans,
label_to_filename=(DATA_PATH, ))
a.fit([['methane', 'mid', 'bad']])
