def sanitizeOEMolecule(molecule):
"""
This function checks if the molecule has coordinates,
explicit hydrogens, aromaticity missing and not unique
atom names. If the molecule does not have coordinates
a fatal error is raised. If the molecule does not have
hydrogens or aramatic flags are missing then a copy of
the molecule is fixed, if missing or not unique atom
names are found then a copy of the molecule is fixed
Parameters:
-----------
molecule: OEMol
The molecule to be checked
Return:
-------
mol_copy: OEMol
A copy of the checked molecule with fixed aromaticity,
hydrogens and unique atom names if they are missing
"""
mol_copy = oechem.OEMol(molecule)
# Check if the molecule has 3D coordinates
if not mol_copy.NumAtoms() == 1: # Mono-atomic molecules are skipped
if not oechem.OEGetDimensionFromCoords(mol_copy):
raise ValueError("The molecule coordinates are set to zero")
# Check if the molecule has hydrogens
if not oechem.OEHasExplicitHydrogens(mol_copy):
oechem.OEAddExplicitHydrogens(mol_copy)
# Check if the molecule has assigned aromaticity
if not mol_copy.HasPerceived(oechem.OEPerceived_Aromaticity):
# oechem.OEAssignAromaticFlags(mol_copy, oechem.OEAroModelOpenEye)
oechem.OEAssignAromaticFlags(mol_copy, oechem.OEAroModelMDL)
if not mol_copy.HasPerceived(oechem.OEPerceived_Chiral):
oechem.OEPerceiveChiral(mol_copy)
# Check for any missing and not unique atom names.
# If found reassign all of them as Tripos atom names
atm_list_names = []
for atom in mol_copy.GetAtoms():
atm_list_names.append(atom.GetName())
reassign_names = False
if len(set(atm_list_names)) != len(atm_list_names):
reassign_names = True
if '' in atm_list_names:
reassign_names = True
if reassign_names:
oechem.OETriposAtomNames(mol_copy)
return mol_copy
def rdmol_from_oemol(oemol):
"""
Create an RDKit molecule identical to the input OpenEye molecule.
Reference
---------
Written by Caitlin Bannan:
https://gist.github.com/bannanc/810ccc4636b930a4522636baab1965a6
May not be needed in newer openforcefield versions. See:
https://github.com/openforcefield/openforcefield/issues/135
Parameters
----------
oemol : OEMol
"""
#print("Starting OpenEye molecule: ", oechem.OEMolToSmiles(oemol))
# start function
rdmol = Chem.RWMol()
# RDKit keeps bond order as a type instead using these values, I don't really understand 7,
# I took them from Shuzhe's example linked above
_bondtypes = {
1: Chem.BondType.SINGLE,
1.5: Chem.BondType.AROMATIC,
2: Chem.BondType.DOUBLE,
3: Chem.BondType.TRIPLE,
4: Chem.BondType.QUADRUPLE,
5: Chem.BondType.QUINTUPLE,
6: Chem.BondType.HEXTUPLE,
7: Chem.BondType.ONEANDAHALF,
}
# atom map lets you find atoms again
map_atoms = dict() # {oe_idx: rd_idx}
for oea in oemol.GetAtoms():
oe_idx = oea.GetIdx()
rda = Chem.Atom(oea.GetAtomicNum())
rda.SetFormalCharge(oea.GetFormalCharge())
rda.SetIsAromatic(oea.IsAromatic())
# unlike OE, RDK lets you set chirality directly
cip = oechem.OEPerceiveCIPStereo(oemol, oea)
if cip == oechem.OECIPAtomStereo_S:
rda.SetChiralTag(Chem.CHI_TETRAHEDRAL_CW)
if cip == oechem.OECIPAtomStereo_R:
rda.SetChiralTag(Chem.CHI_TETRAHEDRAL_CCW)
map_atoms[oe_idx] = rdmol.AddAtom(rda)
# As discussed above, setting bond stereochemistry requires neighboring bonds
# so we will store that information by atom index in this list
stereo_bonds = list()
# stereo_bonds will have tuples with the form (rda1, rda2, rda3, rda4, is_cis)
# where rda[n] is an atom index for a double bond of form 1-2=3-4
# and is_cis is a Boolean is True then onds 1-2 and 3-4 are cis to each other
for oeb in oemol.GetBonds():
# get neighboring rd atoms
rd_a1 = map_atoms[oeb.GetBgnIdx()]
rd_a2 = map_atoms[oeb.GetEndIdx()]
# AddBond returns the total number of bonds, so addbond and then get it
rdmol.AddBond(rd_a1, rd_a2)
rdbond = rdmol.GetBondBetweenAtoms(rd_a1, rd_a2)
# Assign bond type, which is based on order unless it is aromatic
order = oeb.GetOrder()
if oeb.IsAromatic():
rdbond.SetBondType(_bondtypes[1.5])
rdbond.SetIsAromatic(True)
else:
rdbond.SetBondType(_bondtypes[order])
rdbond.SetIsAromatic(False)
# If the bond has specified stereo add the required information to stereo_bonds
if oeb.HasStereoSpecified(oechem.OEBondStereo_CisTrans):
# OpenEye determined stereo based on neighboring atoms so get two outside atoms
n1 = [n for n in oeb.GetBgn().GetAtoms() if n != oeb.GetEnd()][0]
n2 = [n for n in oeb.GetEnd().GetAtoms() if n != oeb.GetBgn()][0]
rd_n1 = map_atoms[n1.GetIdx()]
rd_n2 = map_atoms[n2.GetIdx()]
stereo = oeb.GetStereo([n1, n2], oechem.OEBondStereo_CisTrans)
if stereo == oechem.OEBondStereo_Cis:
print('cis')
stereo_bonds.append((rd_n1, rd_a1, rd_a2, rd_n2, True))
elif stereo == oechem.OEBondStereo_Trans:
print('trans')
stereo_bonds.append((rd_n1, rd_a1, rd_a2, rd_n2, False))
# add bond stereochemistry:
for (rda1, rda2, rda3, rda4, is_cis) in stereo_bonds:
# get neighbor bonds
bond1 = rdmol.GetBondBetweenAtoms(rda1, rda2)
bond2 = rdmol.GetBondBetweenAtoms(rda3, rda4)
# Since this is relative, the first bond always goes up
# as explained above these names come from SMILES slashes so UP/UP is Trans and Up/Down is cis
bond1.SetBondDir(Chem.BondDir.ENDUPRIGHT)
if is_cis:
bond2.SetBondDir(Chem.BondDir.ENDDOWNRIGHT)
else:
bond2.SetBondDir(Chem.BondDir.ENDUPRIGHT)
# if oemol has coordinates (The dimension is non-zero)
# add those coordinates to the rdmol
if oechem.OEGetDimensionFromCoords(oemol) > 0:
conformer = Chem.Conformer()
oecoords = oemol.GetCoords()
for oe_idx, rd_idx in map_atoms.items():
(x, y, z) = oecoords[oe_idx]
conformer.SetAtomPosition(rd_idx, Geometry.Point3D(x, y, z))
rdmol.AddConformer(conformer)
# Save the molecule title
rdmol.SetProp("_Name", oemol.GetTitle())
# Cleanup the rdmol
# Note I copied UpdatePropertyCache and GetSSSR from Shuzhe's code to convert oemol to rdmol here:
rdmol.UpdatePropertyCache(strict=False)
Chem.GetSSSR(rdmol)
# I added AssignStereochemistry which takes the directions of the bond set
# and assigns the stereochemistry tags on the double bonds
Chem.AssignStereochemistry(rdmol, force=False)
#print("Final RDKit molecule: ", Chem.MolToSmiles(Chem.RemoveHs(rdmol), isomericSmiles=True))
return rdmol.GetMol()