def test_optimized_only(self):
mols = list(
smu_utils_lib.conformer_to_molecules(
self.conformer,
include_initial_geometries=False,
include_optimized_geometry=True,
include_all_bond_topologies=False))
self.assertLen(mols, 1)
self.assertEqual(
mols[0].GetProp('_Name'),
'SMU 618451001 bt=618451(0/2) geom=opt',
)
self.assertEqual(
'[H]C(F)=C(OC([H])([H])[H])OC([H])([H])[H]',
Chem.MolToSmiles(mols[0], kekuleSmiles=True, isomericSmiles=False))
# This is just two random atoms I picked from the .dat file and converted to
# angstroms instead of bohr.
self.assertEqual('C', mols[0].GetAtomWithIdx(1).GetSymbol())
np.testing.assert_allclose([0.540254, -3.465543, 3.456982],
list(mols[0].GetConformer().GetAtomPosition(1)),
atol=1e-6)
self.assertEqual('H', mols[0].GetAtomWithIdx(13).GetSymbol())
np.testing.assert_allclose([2.135153, -1.817366, 0.226376],
list(mols[0].GetConformer().GetAtomPosition(13)),
atol=1e-6)
def test_initial_only(self):
mols = list(
smu_utils_lib.conformer_to_molecules(
self.conformer,
include_initial_geometries=True,
include_optimized_geometry=False,
include_all_bond_topologies=False))
self.assertLen(mols, 2)
self.assertEqual([m.GetProp('_Name') for m in mols], [
'SMU 618451001 bt=618451(0/2) geom=init(0/2)',
'SMU 618451001 bt=618451(0/2) geom=init(1/2)',
])
# This is just one random atom I picked from the .dat file and converted to
# angstroms instead of bohr.
self.assertEqual('C', mols[0].GetAtomWithIdx(1).GetSymbol())
np.testing.assert_allclose(
[0.6643, -3.470301, 3.4766],
list(mols[0].GetConformer().GetAtomPosition(1)),
atol=1e-6)
self.assertEqual('C', mols[1].GetAtomWithIdx(1).GetSymbol())
np.testing.assert_allclose(
[664.299998, -3470.300473, 3476.600215],
list(mols[1].GetConformer().GetAtomPosition(1)),
atol=1e-6)
def write(self, conformer):
"""Writes a Conformer.
Args:
conformer: dataset_pb2.Conformer
"""
for mol in smu_utils_lib.conformer_to_molecules(
conformer,
include_initial_geometries=self.init_geometry,
include_optimized_geometry=self.opt_geometry,
include_all_bond_topologies=True):
self.writer.write(mol)
def test_all_outputs(self):
mols = list(smu_utils_lib.conformer_to_molecules(self.conformer))
self.assertLen(mols, 6) # 2 bond topologies * (1 opt geom + 2 init_geom)
self.assertEqual([m.GetProp('_Name') for m in mols], [
'SMU 618451001 bt=618451(0/2) geom=init(0/2)',
'SMU 618451001 bt=618451(0/2) geom=init(1/2)',
'SMU 618451001 bt=618451(0/2) geom=opt',
'SMU 618451001 bt=99999(1/2) geom=init(0/2)',
'SMU 618451001 bt=99999(1/2) geom=init(1/2)',
'SMU 618451001 bt=99999(1/2) geom=opt'
])
self.assertEqual(
'[H]C(F)=C(OC([H])([H])[H])OC([H])([H])[H]',
Chem.MolToSmiles(mols[0], kekuleSmiles=True, isomericSmiles=False))
self.assertEqual(
'[H]C(F)=C(OC([H])([H])[H])OC([H])([H])[H]',
Chem.MolToSmiles(mols[4], kekuleSmiles=True, isomericSmiles=False))
db = smu_sqlite.SMUSQLite('20220104_standard.sqlite', 'r')
# We will look at one conformer that illustrates the complexities of
# converting a conformer to molecules(s). Not all conformers will have
# this level of complexity.
conformer = db.find_by_conformer_id(8240001)
# This RDKit object will be used to write SDF files to stdout to illustrate
# the example.
writer = Chem.SDWriter(sys.stdout)
# We'll start with the simplest case that always generates a single molcule.
case0_mols = list(
smu_utils_lib.conformer_to_molecules(conformer,
include_initial_geometries=False,
include_optimized_geometry=True,
include_all_bond_topologies=False))
assert len(case0_mols) == 1
# Note the three "include" arguments above
# * include_initial_geometries: means to generate molecules for all the
# initial geometries. There will always be at least one initial
# geometry, but there can be many.
# * include_optimized_geometry: means to include output for the
# (single) optimized geometry.
# * include_all_bond_topologies: False means to use only the first
# (i.e. best matching) bond topology that fits this geometry. True
# means to generate separate rdkit molcules for each topology.
print(
'A single molecule comes from asking for only the optimized geometry and only',
