def prepare_mol(self, mol: rdchem.Mol) -> Tuple[str, rdchem.Mol]:
"""Prepare both smiles and mol by standardizing to common rules.
This method should be called before `get_input_feats`.
Params:
-------
mol: rdkit.Chem.rdchem.Mol
Molecule of interest.
Returns:
--------
canonical_smiles: str
Canonical SMILES representation of the molecule.
mol: rdkit.Chem.rdchem.Mol
Modified molecule w/ kekulization and Hs added, if specified.
"""
canonical_smiles = rdmolfiles.MolToSmiles(mol, canonical=True)
mol = rdmolfiles.MolFromSmiles(canonical_smiles)
if self.add_Hs:
mol = rdmolops.AddHs(mol)
if self.kekulize:
rdmolops.Kekulize(mol)
return canonical_smiles, mol
def process_molecule(self, pdb_file, use_esp=False):
"""
Processes a molecule from the passed PDB file if the file contents has
no errors.
:param pdb_file: path to the PDB file to process the molecule from.
:return: a ProcessedMolecule object
"""
# NOTE: Gasteiger is an inappropriate algorithm for ESP calculation of proteins!
# read a molecule from the PDB file
try:
mol = Chem.MolFromPDBFile(molFileName=pdb_file, removeHs=False,
sanitize=True)
except IOError:
log.warning("Could not read PDB file.")
return None
if mol is None:
log.warning("Bad pdb file found.")
return None
if use_esp:
try:
# add missing hydrogen atoms
mol = rdMO.AddHs(mol, addCoords=True)
# compute partial charges
rdPC.ComputeGasteigerCharges(mol, throwOnParamFailure=True)
except ValueError:
log.warning("Bad Gasteiger charge evaluation.")
return None
# get the conformation of the molecule
conformer = mol.GetConformer()
# calculate the center of the molecule
center = rdMT.ComputeCentroid(conformer, ignoreHs=False)
atoms_count = mol.GetNumAtoms()
atoms = mol.GetAtoms()
def get_coords(i):
coord = conformer.GetAtomPosition(i)
return np.asarray([coord.x, coord.y, coord.z])
# set the coordinates, charges, VDW radii and atom count
res = {
"coords": np.asarray(
[get_coords(i) for i in range(0, atoms_count)]) - np.asarray(
[center.x, center.y, center.z]),
"vdwradii": np.asarray(
[self.periodic_table.GetRvdw(atom.GetAtomicNum()) for atom in
atoms])
}
if use_esp:
res['charges'] = np.asarray([float(atom.GetProp("_GasteigerCharge")) for atom in atoms])
return res
def get_all_interactions(self, pdbfile, sdffile):
residue_names, residue_atoms = get_residues_and_atoms(pdbfile)
compounds = get_compounds(sdffile)
all_interactions = {}
for index, compound in enumerate(compounds):
compound = rdmolops.AddHs(compound)
interactions = self.get_interactions(compound, residue_atoms,
residue_names)
all_interactions[index] = interactions
return all_interactions
def test_mols():
mols = []
all_smiles = [
'CN=C=O', 'Cc1ccccc1', 'CC1=CC2CC(CC1)O2', 'CCCCCCCCCCCCCCCC'
]
for smiles in all_smiles:
mol = rdmolfiles.MolFromSmiles(smiles)
mol = rdmolops.AddHs(mol, addCoords=True)
rdDistGeom.EmbedMolecule(mol, rdDistGeom.ETKDG())
mol = rdmolops.RemoveHs(mol)
mol.SetProp('Fitness', str(np.random.rand(1)[0]))
mols.append(mol)
return mols
def apply_retrorules(self, smile, rxns, explicit_hydrogens=False):
'''Function takes a smile and dictionary of reactions, applys the reactions and
returns a dictionary of rxn_names : products '''
try:
substrate_molecule = AllChem.MolFromSmiles(smile)
except:
return {}
if explicit_hydrogens == True:
substrate_molecule = rdmolops.AddHs(substrate_molecule)
rxn_product_dict = {}
for rxn_name, rxn in rxns.items():
try:
products = rxn.RunReactants((substrate_molecule, ))
except:
products = []
print('Error running reactants for: ' + str(smile))
smiles_products = []
for product in products:
sub_list = []
for mol in product:
mols = [mol]
if explicit_hydrogens == True:
mol = rdmolops.RemoveHs(mol)
try:
mols = rdmolops.GetMolFrags(mol, asMols=True)
except:
pass
for mol in mols:
try:
p_smile = AllChem.MolToSmiles(mol)
p_smile = rdkit_smile(p_smile)
if self._check_valid_smile(
p_smile, rxn_name=rxn_name) == True:
sub_list.append(p_smile)
except:
pass
if (sub_list not in smiles_products) and (len(sub_list) != 0):
smiles_products.append(sub_list)
if len(smiles_products) != 0:
rxn_product_dict[rxn_name] = smiles_products
return rxn_product_dict
def opt_geometry(mol, max_iter, mmffvariant, seed, max_attempts):
err = 0
try:
mol = rdmolops.AddHs(mol)
a = AllChem.EmbedMolecule(mol,
useRandomCoords=True,
useBasicKnowledge=True,
randomSeed=seed,
clearConfs=True,
maxAttempts=max_attempts)
if a == -1:
err = 0
AllChem.MMFFOptimizeMolecule(mol,
maxIters=max_iter,
mmffVariant=mmffvariant)
except ValueError:
err = 1
except TypeError:
err = 1
return mol, err
def construct_pos_matrix(mol: rdchem.Mol,
out_size: Optional[int] = -1) -> np.ndarray:
"""Construct relative positions from each atom within the molecule.
Params:
-------
mol: rdkit.Chem.rdchem.Mol
Molecule of interest.
out_size: int, optional, default=-1
The size of the returned array. If this option is negative, it
does not take any effect. Otherwise, it must be larger than or
equal to the number of atoms in the input molecule. If so, the
end of the array is padded with zeros.
Returns:
--------
pos_matrix: np.ndarray, shape=(n,n,3)
Relative position (XYZ) coordinates from one atom the others in
the mol.
Examples:
---------
```python
>>> from rdkit import Chem
>>> from rdkit.Chem import AllChem
>>> smiles = 'N[C@@]([H])([C@]([H])(O2)C)C(=O)N[C@@]([H])(CC(=O)N)C(=O)N[C@@]([H])([C@]([H])' \
'(O)C)C(=O)N[C@@]([H])(Cc1ccc(O)cc1)C(=O)2'
>>> mol = Chem.MolFromSmiles(smiles)
>>> mol = Chem.AddHs(mol, addCoords=True)
>>> AllChem.EmbedMolecule(mol, AllChem.ETKDG())
>>> mol = Chem.RemoveHs(mol)
>>> pos_matrix = construct_pos_matrix(mol, out_size=-1)
>>> pos_matrix.shape
(34,34,3)
>>> pos_matrix = construct_pos_matrix(mol, out_size=49)
>>> pos_matrix.shape
(49,49,3)
```
"""
# Obtain initial distance geometry between atoms, if unavilable
if mol.GetNumConformers() == 0:
mol = rdmolops.AddHs(mol, addCoords=True)
rdDistGeom.EmbedMolecule(mol, rdDistGeom.ETKDG())
mol = rdmolops.RemoveHs(mol)
coords = mol.GetConformer().GetPositions() # shape=(N,3)
N = mol.GetNumAtoms()
# Determine appropiate output size to generate feature matrix of same size for all mols.
if out_size < 0:
size = N
elif out_size >= N:
size = out_size
else:
raise ValueError(
'`out_size` (N={}) is smaller than number of atoms in mol (N={})'.
format(out_size, N))
pos_matrix = np.zeros(shape=(size, size, 3), dtype=np.float)
for atom_idx in range(N):
atom_pos = coords[atom_idx] # central atom of interest
for neighbor_idx in range(N):
neigh_pos = coords[neighbor_idx] # neighboring atom
pos_matrix[
atom_idx,
neighbor_idx] = atom_pos - neigh_pos # dist between neighbor -> center
return pos_matrix
def sample_mol():
mol = rdmolfiles.MolFromSmiles('CN=C=O')
mol = rdmolops.AddHs(mol, addCoords=True)
rdDistGeom.EmbedMolecule(mol, rdDistGeom.ETKDG())
return rdmolops.RemoveHs(mol)
def identify_functional_groups(smi):
## We decided to start from a SMILES and add explicit hydrogens inside the function
mol = Chem.MolFromSmiles(smi)
mol = rdmolops.AddHs(mol)
try:
marked = set()
## Since heteroatoms are included in PATT_TUPLE, we remove the first part of the original function
for patt in PATT_TUPLE:
for path in mol.GetSubstructMatches(patt):
for atomindex in path:
marked.add(atomindex)
#merge all connected marked atoms to a single FG
groups = []
while marked:
grp = set([marked.pop()])
merge(mol, marked, grp)
groups.append(grp)
groups = [list(x) for x in groups]
## It seems that the initial filtering of heteroatoms was not enough, so we add this to remove groups with only aromatic atoms
for g in groups:
group_aromaticity = set(
[mol.GetAtomWithIdx(idx).GetIsAromatic() for idx in g])
if group_aromaticity == {True}:
groups.remove(g)
## Identify bonds to break and hydrogens to keep for every FG
bonds = []
labels = []
for g in groups:
group_bonds = []
group_labels = []
for idx in g:
atom = mol.GetAtomWithIdx(idx)
## Carbon atoms
if atom.GetAtomicNum() == 6:
for nbr in atom.GetNeighbors():
## Carbonyl groups to disciminate between aldehydes and ketones
if nbr.GetAtomicNum() == 8 and str(
mol.GetBondBetweenAtoms(
idx,
nbr.GetIdx()).GetBondType()) == "DOUBLE":
PreserveH = True
break
else:
PreserveH = False
if PreserveH == True:
for nbr in atom.GetNeighbors():
jdx = nbr.GetIdx()
if jdx not in g and nbr.GetAtomicNum() != 1:
group_bonds.append(
mol.GetBondBetweenAtoms(idx, jdx).GetIdx())
group_labels.append((0, 0))
else:
for nbr in atom.GetNeighbors():
jdx = nbr.GetIdx()
if jdx not in g:
group_bonds.append(
mol.GetBondBetweenAtoms(idx, jdx).GetIdx())
group_labels.append((0, 0))
## Nitrogen atoms
elif atom.GetAtomicNum() == 7:
## To discriminate between anilines and amines (primary, secondary, etc)
if len(g) == 1:
neigh_atn = [
x.GetAtomicNum() for x in atom.GetNeighbors()
if x.GetAtomicNum() != 1
]
if neigh_atn.count(6) == 1:
for nbr in atom.GetNeighbors():
jdx = nbr.GetIdx()
if jdx not in g and nbr.GetAtomicNum() != 1:
group_bonds.append(
mol.GetBondBetweenAtoms(idx,
jdx).GetIdx())
if nbr.GetIsAromatic() == True:
group_labels.append((1, 1))
else:
group_labels.append((0, 0))
else:
for nbr in atom.GetNeighbors():
jdx = nbr.GetIdx()
if jdx not in g and nbr.GetAtomicNum() != 1:
group_bonds.append(
mol.GetBondBetweenAtoms(idx,
jdx).GetIdx())
group_labels.append((0, 0))
else:
for nbr in atom.GetNeighbors():
jdx = nbr.GetIdx()
if jdx not in g:
group_bonds.append(
mol.GetBondBetweenAtoms(idx, jdx).GetIdx())
group_labels.append((0, 0))
## Oxygen atoms
elif atom.GetAtomicNum() == 8:
## To discriminate between alcohols from phenols and esthers from carboxylic acids
if len(g) == 1:
neigh_atn = [
x.GetAtomicNum() for x in atom.GetNeighbors()
if x.GetAtomicNum() != 1
]
if len(neigh_atn) == 1 and neigh_atn.count(6) == 1:
for nbr in atom.GetNeighbors():
jdx = nbr.GetIdx()
if jdx not in g and (nbr.GetAtomicNum() != 1):
group_bonds.append(
mol.GetBondBetweenAtoms(idx,
jdx).GetIdx())
if nbr.GetIsAromatic() == True:
group_labels.append((1, 1))
else:
group_labels.append((0, 0))
else:
for nbr in atom.GetNeighbors():
jdx = nbr.GetIdx()
if jdx not in g and nbr.GetAtomicNum() != 1:
group_bonds.append(
mol.GetBondBetweenAtoms(idx,
jdx).GetIdx())
group_labels.append((0, 0))
else:
for nbr in atom.GetNeighbors():
jdx = nbr.GetIdx()
if jdx not in g and nbr.GetAtomicNum() != 1:
group_bonds.append(
mol.GetBondBetweenAtoms(idx, jdx).GetIdx())
group_labels.append((0, 0))
## Sulfur atoms
elif atom.GetAtomicNum() == 16:
if len(g) == 1:
for nbr in atom.GetNeighbors():
jdx = nbr.GetIdx()
if jdx not in g and nbr.GetAtomicNum() != 1:
group_bonds.append(
mol.GetBondBetweenAtoms(idx, jdx).GetIdx())
group_labels.append((0, 0))
else:
for nbr in atom.GetNeighbors():
jdx = nbr.GetIdx()
if jdx not in g:
group_bonds.append(
mol.GetBondBetweenAtoms(idx, jdx).GetIdx())
group_labels.append((0, 0))
else:
for nbr in atom.GetNeighbors():
jdx = nbr.GetIdx()
if jdx not in g:
group_bonds.append(
mol.GetBondBetweenAtoms(idx, jdx).GetIdx())
group_labels.append((0, 0))
labels.append(group_labels)
bonds.append(group_bonds)
## Build final fragments
FGS_ENVS = []
for i in range(len(groups)):
Frag = Chem.FragmentOnBonds(mol, bonds[i], dummyLabels=labels[i])
Frags = rdmolops.GetMolFrags(Frag)
for j in Frags:
if groups[i][0] in j:
FGS_ENVS.append(
Chem.MolFragmentToSmiles(Frag,
j,
canonical=True,
allHsExplicit=True))
FGS_ENVS = list(set(FGS_ENVS))
for i in FGS_ENVS:
if Chem.MolFromSmiles(i) == None:
FG = Chem.MolFromSmarts(i)
else:
FG = Chem.MolFromSmiles(i)
if set([
atom.GetIsAromatic() for atom in FG.GetAtoms()
if atom.GetSymbol() not in ["*", "H"]
]) == {True}:
FGS_ENVS.remove(i)
return FGS_ENVS
except:
## When the molecules is as small as a single FG
FGS_ENVS = [Chem.MolToSmiles(mol, canonical=True, allHsExplicit=True)]
return FGS_ENVS
def process_molecule(self, pdb_file):
"""
Splits the molecules into separate channels.
:param pdb_file: the pdb file to be processed
:return: a dictionary of the coordinates and vdwradii for each channel
"""
hydro_file_name = '_hydrogenized.'.join(
os.path.basename(pdb_file).split('.'))
hydrogenized_pdb_file = os.path.join(os.path.dirname(pdb_file),
hydro_file_name)
try:
mol_rdkit = Chem.MolFromPDBFile(molFileName=pdb_file,
removeHs=False, sanitize=True)
if mol_rdkit is not None:
mol_rdkit = rdMO.AddHs(mol_rdkit, addCoords=True)
# get the conformation of the molecule
conformer = mol_rdkit.GetConformer()
# calculate the center of the molecule
center = rdMT.ComputeCentroid(conformer, ignoreHs=False)
mol_center = np.asarray([center.x, center.y, center.z])
else:
raise ValueError
pdbw = Chem.rdmolfiles.PDBWriter(fileName=hydrogenized_pdb_file)
pdbw.write(mol_rdkit)
pdbw.flush()
pdbw.close()
del mol_rdkit, pdbw
except (IOError, ValueError):
log.warning("Bad PDB file.")
return None
try:
mol = pd.parsePDB(hydrogenized_pdb_file)
except IOError:
log.warning("Could not read PDB file.")
return None
if mol is None:
log.warning("Bad pdb file found.")
return None
std_amino_acids = ['ALA', 'ARG', 'ASN', 'ASP', 'CYS',
'GLN', 'GLU', 'GLY', 'HIS', 'ILE',
'LEU', 'LYS', 'MET', 'PHE', 'PRO',
'SER', 'THR', 'TRP', 'TYR', 'VAL']
canonical_notation = lambda x: x[0].upper() + x[1:].lower() if len(
x) > 1 else x
res = {'coords': mol.getCoords() - mol_center,
'vdwradii': np.asarray([self.periodic_table.GetRvdw(
self.periodic_table.GetAtomicNumber(
canonical_notation(atom)))
for atom in mol.getElements()])}
# find the data for all the 20 amino acids
for aa in std_amino_acids:
all_aas_in_mol = mol.select('resname ' + aa)
if all_aas_in_mol is not None:
mask = all_aas_in_mol.getIndices()
else:
mask = np.array([], dtype=np.int32)
res['coords_' + aa] = res['coords'][mask, :]
res['vdwradii_' + aa] = res['vdwradii'][mask]
# find the data for the backbones
backbone_mask = mol.backbone.getIndices()
res['coords_backbone'] = res['coords'][backbone_mask, :]
res['vdwradii_backbone'] = res['vdwradii'][backbone_mask]
# find the data for the heavy atoms (i.e. no H atoms)
heavy_mask = mol.heavy.getIndices()
res['coords_heavy'] = res['coords'][heavy_mask, :]
res['vdwradii_heavy'] = res['vdwradii'][heavy_mask]
# find the data for the heavy atoms (i.e. no H atoms)
hydro_mask = mol.hydrogen.getIndices()
res['coords_hydro'] = res['coords'][hydro_mask, :]
res['vdwradii_hydro'] = res['vdwradii'][hydro_mask]
return res