You are assuming that CHI_TETRAHEDRAL_CW means "R" and v.v. for "S". I don't think that's correct.
On 8 September 2015 at 12:58, Michał Nowotka <[email protected]> wrote: > Hi, > > I'm comparing stereo centers assigned by ChemAxon JChem and RDKit. > > The code would be: > > #---------------------------------------------------------------------------------------------------------------------- > > from rdkit import Chem > > def getStereoInfo(smiles): > ret = [] > mol = Chem.MolFromSmiles(smiles) > Chem.AssignStereochemistry(mol, flagPossibleStereoCenters=True, > force=True) > for atom in mol.GetAtoms(): > stereo = str(atom.GetChiralTag()) > atomIndex = atom.GetIdx() > if str(atom.GetChiralTag()) != "CHI_UNSPECIFIED": > if stereo == "CHI_TETRAHEDRAL_CW": > chirality = "R" > elif stereo == "CHI_TETRAHEDRAL_CCW": > chirality = "S" > else: > chirality = "R/S" > ret.append({"atomIndex":atomIndex,"chirality":chirality}) > return ret > > #------------------------------------------------------------------------------------------------------------------- > > Now for (1R,2S,3S)-1,2,3-trichlorocyclopentane > (Cl[C@H]1CC[C@@H](Cl)[C@H]1Cl), both RDKit and JChem give the same > results: > https://dl.dropboxusercontent.com/u/10967207/a.png > > But for ephedrine (O[C@H](c1ccccc1)[C@@H](NC)C) stereo centers are > reversed. RDKit gives: > https://dl.dropboxusercontent.com/u/10967207/b.png > > and JChem gives: > https://dl.dropboxusercontent.com/u/10967207/c.png > > The same is shown by pseudo-ephedrine (O[C@@H](c1ccccc1)[C@@H](NC)C) > which returns both stereo centers as 'R'. > > Is that a bug or an expected behavior? Are there any parameters in > RDKit that can be set to get results similar to JChem? > > Regards, > > Michał Nowotka > > ------------------------------------------------------------------------------ > _______________________________________________ > Rdkit-discuss mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/rdkit-discuss ------------------------------------------------------------------------------ _______________________________________________ Rdkit-discuss mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
