mfp_symmetrization.py

#https://github.com/rdkit/rdkit/discussions/6135

from rdkit import Chem
from rdkit.Chem import AllChem

def mol_to_bits(mol, radius=0):
    bi = {}
    fp = AllChem.GetMorganFingerprint(mol, radius, bitInfo=bi,useFeatures=False)
    bits = np.zeros(mol.GetNumAtoms(),dtype=int)
    for key, value in bi.items():
        for v in value:
            if v[1]==radius:
                bits[v[0]] = key
    return bits

m = Chem.MolFromSmiles('C([O-])=O')
print('0th neighbourhood bits:', mol_to_bits(m, radius=1))

substruc = Chem.MolFromSmarts('[C](-O)=O')

print(m.GetSubstructMatches(substruc))

#for radius=1, also set bonds to be symmetrical
bond = m.GetBondBetweenAtoms(0, 1)
bond.SetBondType(Chem.BondType.ONEANDAHALF)
bond = m.GetBondBetweenAtoms(0, 2)
bond.SetBondType(Chem.BondType.ONEANDAHALF)

m.GetAtomWithIdx(1).SetFormalCharge(-1)
m.GetAtomWithIdx(2).SetFormalCharge(-1)

#note, two oxygens are equal now.
mol_to_bits(m, radius=1)