Stereoisomers for achiral molecules are generated

[DrrDom]

Loks as a bug of RDKit (rdkit/rdkit#8595), which affects the preparation pipeline. It creates a wedged bonds for centers which are not chiral bridgeheads in symmetric bicyclic systems. This does not affect docking, but unnecessary change the input SMILES, that may cause inconsistencies at later stages of the analysis.

from rdkit import Chem
from rdkit.Chem.EnumerateStereoisomers import EnumerateStereoisomers, StereoEnumerationOptions

smi = 'NC(=O)[C@@H]1CC2CC1C2'
mol = Chem.MolFromSmiles(smi)

opts = StereoEnumerationOptions(tryEmbedding=True, maxIsomers=10)
isomers = tuple(EnumerateStereoisomers(mol, options=opts))
print(smi)
for m in isomers:
    print(Chem.MolToSmiles(m))

output

NC(=O)[C@@H]1CC2CC1C2  # source smi
NC(=O)[C@@H]1C[C@H]2C[C@@H]1C2  # output smi

A workaround should be implemented for this issue.