def pka_similarities(smile, mol_set, n):
mol = Chem.MolFromSmiles(smile)
mol_fp = rdMolDescriptors.GetHashedAtomPairFingerprintAsBitVect(mol)
similarity = []
for molecule in mol_set:
sim = DataStructs.DiceSimilarity(mol_fp, molecule[2])
similarity.append([sim, molecule[1]])
return np.asarray(sorted(similarity)[:n]).flatten()
def _filter_for_similar_ligands_2d(
ligand: oechem.OEGraphMol,
structures: pd.DataFrame) -> pd.DataFrame:
"""
Filter KLIFS structures for similar ligands according to a fingerprint comparison.
Parameters
----------
ligand: oechem.OEGraphMol
An OpenEye molecule holding the ligand to dock.
structures: pd.DataFrame
A DataFrame containing KLIFS entries.
Returns
-------
: pd.DataFrame
The input DataFrame filtered for KLIFS entries with most similar ligands.
"""
import pandas as pd
from openeye import oechem
from rdkit import Chem, RDLogger
from rdkit.Chem import AllChem, DataStructs
RDLogger.DisableLog("rdApp.*") # disable RDKit logging
logging.debug("Converting OpenEye molecule to RDKit molecule ...")
ligand = Chem.MolFromSmiles(oechem.OEMolToSmiles(ligand))
logging.debug("Converting SMILES to RDKit molecules ...")
rdkit_molecules = [
Chem.MolFromSmiles(smiles) for smiles in structures.smiles
]
logging.debug("Adding RDKit molecules to dataframe...")
structures["rdkit_molecules"] = rdkit_molecules
logging.debug(
"Removing KLIFS entries without valid RDKit molecule ...")
structures = structures[structures.rdkit_molecules.notnull()]
logging.debug("Adding Feature Morgan fingerprint to dataframe...")
pd.options.mode.chained_assignment = None # otherwise next line would raise a warning
structures["rdkit_fingerprint"] = [
AllChem.GetMorganFingerprint(rdkit_molecule, 2, useFeatures=True)
for rdkit_molecule in structures.rdkit_molecules
]
logging.debug("Generating Feature Morgan fingerprint of ligand ...")
ligand_fingerprint = AllChem.GetMorganFingerprint(ligand,
2,
useFeatures=True)
logging.debug("Calculating dice similarity between fingerprints ...")
fingerprint_similarities = [[
i, DataStructs.DiceSimilarity(ligand_fingerprint, fingerprint)
] for i, fingerprint in enumerate(structures.rdkit_fingerprint)]
# if maximal score is 0.87, threshold is set to 0.77
fingerprint_similarity_threshold = (
max([similarity[1]
for similarity in fingerprint_similarities]) - 0.1)
logging.debug("Picking structures with most similar ligands ...")
structures = structures.iloc[[
similarity[0] for similarity in fingerprint_similarities
if similarity[1] >= fingerprint_similarity_threshold
]]
return structures
def __call__(self, a, b):
return 1 - DataStructs.DiceSimilarity(a, b)
