Hello, very interesting. I think it is worth raising a related point regarding the representation of stereochemistry for such types of molecules, because it’s important not only to validate them, but to interpret them correctly, too. See this document: http://publications.iupac.org/pac/2006/pdf/7810x1897.pdf Page 1926, at the bottom, shows this:
[cid:image006.png@01D7B3AF.AB916440] The question is: when you submit to rdkit a molecule with the wedge pointing to the exocyclic H, is the stereochemistry correctly interpreted by rdkit (regardless of the conformational possibility of it existing or not)? Using Biovia Draw, with the various possibilities (I changed a bit the molecule in the OP to make the CIP assignment more obvious), you see that indeed it’s not as easy as it seems: [cid:image007.jpg@01D7B3B1.6B11C930] If the 6-membered ring bonds are all on the plane, what does an up wedge to the top H atom mean? Presumably, that the bond ‘in the middle’ of the ring, the one going from C to NH, is below the plane. In that case, the stereochemistry for the top C is R. And as you can see, the software ‘gets it wrong’ when using the exocyclic wedge. Perhaps it would be interesting to know what rdkit makes of a molecule like the first one on the left. Does it ‘read’ the stereochemistry correctly? brg Giovanni From: Tim Dudgeon <tdudgeon...@gmail.com> Sent: 27 September 2021 14:04 To: RDKit Discuss <rdkit-discuss@lists.sourceforge.net> Subject: [Rdkit-discuss] validating stereochemistry *** CAUTION : External e-mail *** I have Python code to enumerate undefined chiral centres in a molecule. Mostly this works fine, but for some constrained structures this can generate stereochemistry that makes no sense. For instance consider NC1CC2CCC1C2: [cid:image001.png@01D7B3AF.4C33C970] These two make sense: [cid:image002.png@01D7B3AF.4C33C970] [cid:image003.png@01D7B3AF.4C33C970] But these two don't: [cid:image004.png@01D7B3AF.4C33C970] [cid:image005.png@01D7B3AF.4C33C970] Is there a way to filter out the invalid ones? Thanks Tim
_______________________________________________ Rdkit-discuss mailing list Rdkit-discuss@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
