RDkit One Component Rxn Bug


I have noticed a bug in the RDKit One Component Reaction node.

If you take the reaction transformation of a Pyrazine structure, and replace both the Nitrogens of the Pyrazine for a carbon with a pendant Fluoro (thus a Difluorobenzene), the output of the node gives duplicated structures, two product structures per reactant.

I can see why it is giving two structures, as there replacements groups can be applied to either Nitrogen (i.e. if the replacement was a Fluoro for one carbon and Chloro for the other carbon then yes, two product structures are possible), but in these cases the algorithm is not recognising that the two possible product structures are identical.



You've diagnosed the situation perfectly: the node currently does no duplicate checking on the result sets.

This is a feature that we could add, but it's going to slow the calculations down since it would require generating canonical SMILES for every product, so it will remain an optional feature.




Okay, thanks for the update.


This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.