Return original molecule from StereoisomerEnumerator when no unspecified stereocenters are available

[phonglam3103]

What does this implement/fix? Explain your changes.

The current implementation of rdEnumerateStereoisomers with the tryEmbedding option enabled attempts to embed the input molecule even when no stereoisomer flipping has occurred.

For example:

from rdkit import Chem
from rdkit.Chem import rdEnumerateStereoisomers


mol = Chem.MolFromSmiles(r'C1C\C=C\CCC[C@H]1C(=O)Nc2ccc(cc2)C[C@@H](C(=O)[O-])[NH3+]')
smiles = set()
opts = rdEnumerateStereoisomers.StereoEnumerationOptions()
opts.tryEmbedding=True
enum = rdEnumerateStereoisomers.StereoisomerEnumerator(mol, opts)
while True:
    iso = enum.next()
    if iso is None:
        break
    smiles.add(Chem.MolToSmiles(iso))
print(len(smiles), 'stereoisomers found:')
for smi in smiles:
    print(smi)

returns:

0 stereoisomers found:
[08:22:18] [NH3+][C@@H](Cc1ccc(NC(=O)[C@H]2CC/C=C/CCC2)cc1)C(=O)[O-]     failed to embed.

As a result, the input molecule disappears after enumeration.

By contrast, the original Python implementation of EnumerateStereoisomers returns the input molecule unchanged when no “flippers” (stereocenters to invert) are present:

from rdkit import Chem
from rdkit.Chem import AllChem

opts = AllChem.StereoEnumerationOptions(tryEmbedding=True)
mol = Chem.MolFromSmiles(r'C1C\C=C\CCC[C@H]1C(=O)Nc2ccc(cc2)C[C@@H](C(=O)[O-])[NH3+]')
isomers = tuple(AllChem.EnumerateStereoisomers(mol, options=opts))
print(f'Found {len(isomers)} stereoisomers:')
for iso in isomers:
    print(Chem.MolToSmiles(iso))

returns:

Found 1 stereoisomers:
[NH3+][C@@H](Cc1ccc(NC(=O)[C@H]2CC/C=C/CCC2)cc1)C(=O)[O-]

This PR updates the C++ StereoisomerEnumerator so that it behaves consistently with the Python version. Specifically, it skips embedding when the d_flippers vector is empty (i.e., no new stereoisomers are generated). This ensures that the input molecule is preserved and avoids unnecessary embedding attempts, improving both correctness and performance.

[phonglam3103]

Hi @DavidACosgrove, it would be nice if you can have a look on this PR

[DavidACosgrove]

LGTM.

[DavidACosgrove]

Hi @DavidACosgrove, it would be nice if you can have a look on this PR

Sorry, that I missed that one. One of the maintainers will have to review it also, however, before it can be accepted and merged.

[greglandrum]

LGTM

[greglandrum]

Thanks @phonglam3103 !