def structure(self) -> mtr.Structure:
coords = self.cclib_out.atomcoords[-1, :, :] * mtr.angstrom
zs = self.cclib_out.atomnos
atoms = (mtr.Atom(element=Z, position=p) for Z, p in zip(zs, coords))
return mtr.Structure(*atoms)
def _structure_from_pubchem_compound(compound: pcp.Compound) -> mtr.Structure:
# FIXME: assumes the pubchem distance unit is angstrom - is this correct??
atom_generator = ((a.element, (a.x, a.y, a.z) * mtr.angstrom)
for a in compound.atoms)
atoms = (mtr.Atom(element=symb, position=pos)
for symb, pos in atom_generator)
return mtr.Structure(*atoms)
def _structure_from_identifier(smiles: Optional[str] = None,
inchi: Optional[str] = None,
num_conformers: int = 25) -> mtr.Structure:
# for motivation on generating 25 (as opposed to, say, 10 or 100) conformers, see:
# https://github.com/rdkit/UGM_2015/blob/master/Presentations/ETKDG.SereinaRiniker.pdf
rdkit.RDLogger.DisableLog("rdApp.*")
if smiles is not None:
mol = rdkit.Chem.MolFromSmiles(smiles, sanitize=False)
elif inchi is not None:
mol = rdkit.Chem.MolFromInchi(inchi, sanitize=False)
else:
raise ValueError(
"Either SMILES or InChi required to generate structure.")
# sanitize
try:
mol.UpdatePropertyCache(False)
mol = rdkit.Chem.Mol(mol.ToBinary())
rdkit.Chem.SanitizeMol(mol)
except ValueError:
raise ValueError("Cannot sanitize RDKit molecule.")
# hydrogenate
mol = rdkit.Chem.AddHs(mol)
# embed to generate 3D coords
embedding_parameters = rdkit.Chem.AllChem.ETKDG()
embed_return_code = rdkit.Chem.AllChem.EmbedMolecule(
mol=mol, params=embedding_parameters)
if embed_return_code == -1:
embedding_parameters.useRandomCoords = True
rdkit.Chem.AllChem.EmbedMolecule(mol=mol, params=embedding_parameters)
# embed multiple conformers and find one with lowest energy
rdkit.Chem.AllChem.EmbedMultipleConfs(mol,
numConfs=num_conformers,
params=embedding_parameters)
# MMFF seems to give slightly better geometries, so it is preferred if possible
if rdkit.Chem.AllChem.MMFFHasAllMoleculeParams(mol=mol):
mmff_props = rdkit.Chem.AllChem.MMFFGetMoleculeProperties(mol=mol)
rdkit.Chem.AllChem.MMFFSanitizeMolecule(mol=mol)
def energy(conformer):
return rdkit.Chem.AllChem.MMFFGetMoleculeForceField(
mol=conformer.GetOwningMol(),
pyMMFFMolProperties=mmff_props,
confId=conformer.GetId(),
).CalcEnergy()
else:
def energy(conformer):
return rdkit.Chem.AllChem.UFFGetMoleculeForceField(
mol=conformer.GetOwningMol(),
confId=conformer.GetId()).CalcEnergy()
energies = {
conformer: energy(conformer)
for conformer in mol.GetConformers()
}
conformer = min(energies, key=energies.get)
# convert to Structure
symbols = (a.GetSymbol() for a in conformer.GetOwningMol().GetAtoms())
# FIXME: assumes the RDKIT distance unit is angstrom - is this correct??
# NOTE: using conformer.GetPositions sometimes causes
# a seg fault (RDKit) - use GetAtomPosition instead
atoms = (mtr.Atom(
element=symbol,
position=conformer.GetAtomPosition(i) * mtr.angstrom,
) for i, symbol in enumerate(symbols))
return mtr.Structure(*atoms)
def __add__(self, other: mtr.Structure) -> mtr.Structure:
return mtr.Structure((*self.atoms, *other.atoms))
import materia as mtr
amm = mtr.Structure.read("~/ammonia.xyz")
basis = ase.Atoms(amm.atomic_symbols, amm.atomic_positions.T)
ase_crystal = ase.spacegroup.crystal(
basis=basis,
spacegroup=198,
cellpar=[
4.9621636999999996, 4.9621636999999996, 4.9621636999999996, 90, 90, 90
],
size=(3, 3, 3),
)
ase.io.write("~/ammonia_crystal.xyz", ase_crystal)
crystal = mtr.Structure.read("~/ammonia_crystal.xyz")
atoms = tuple(crystal.atoms[k] for k, v in crystal.perceive_bonds().items()
if (crystal.atoms[k].Z == 7 and len(v) == 3) or (
crystal.atoms[k].Z == 1 and len(v) == 1))
bonds = crystal.perceive_bonds()
while len(bonds) > len(atoms):
bonds = crystal.perceive_bonds()
atoms = tuple(crystal.atoms[k]
for k, v in crystal.perceive_bonds().items()
if (crystal.atoms[k].Z == 7 and len(v) == 3) or (
crystal.atoms[k].Z == 1 and len(v) == 1))
crystal = mtr.Structure(atoms)
crystal.write("~/good_ammonia_crystal.xyz", overwrite=True)