Noel is there a way to represent correlated uncertainty in chirality? When there is only one chiral center, not specifying the chirality is a way to refer to a racemic mixture. But when there are two chiral centers is there a way to represent a racemic mixture of R,R and S,S (and no R,S and S,R) with only one SMILES string? This has been coming up frequently in the melting point literature. - Jean-Claude Bradley

Chirality is specifed on a per atom basis, and so you could say that a particular atom is chiral, only you're not sure of the absolute configuration: C[C@?](Br)(Cl)I. It sounds like you are talking about a mixture of two different molecules. SMILES has a dot disconnected notation for unconnected components in a 1.1 mixture, so you would have MoleculeA.MoleculeB where A and B have opposite stereo specified. - Noel O'Boyle

So a question mark is an actual SMILES character? - Jean-Claude Bradley

Uh, actually, maybe not. Just been looking back through my emails and at opensmiles.org, and it seems that that was just in the devel code. - Noel O'Boyle

OK. I guess the closest we can come is to put the two SMILES connected by a dot. - Jean-Claude Bradley

I regard it as dangerous to use SMILES linked by dots for the representation of racemic mixtures - for the simple reason that even simple property calculations like molecular weight have a wrong result unless you mark the dot-linked SMILES explicitly as racemic mixture in some way (but if you do this you have to add something to the SMILES outside the SMILES anyway). - Markus Sitzmann

Markus - you are right linking enantiomer pairs with a dot isn't exactly the right concept since the molecular weight would be of the entire "complex". The dot makes sense for salts. - Jean-Claude Bradley