Is there an easy way to use reactant identifiers supplied with the reagent structures to index the products, instead of the standard ‘0_0’ etc.? So if I supply say A01, A02 etc. for reactant 1 and B01, B02 for reactant 2, the products would be named A01_B01, A01_B02, etc.
Grateful for any tips!
This is not currently possible directly from the node. It is possible to make it work by adding a few additional nodes to the workflow, but that solution is pretty ugly. If you’re interested I can post an example.
I’ve created a feature request for the RDKit reaction nodes themselves to enable this option and will try to get it into the next release of the nodes.
Being able to select a provided column for each reactant as index would be a nice enhancement to the node.
I can imagine a workaround involving extensive looping right now, but maybe you had something else in mind, so I would certainly appreciate to see your method.
I managed to do a workaround by adding numeric index columns to each list with reactants, several joins, and a Java snippet to generate the product indices. Bit cumbersome admittedly, so having this option built into the RDKit node would be nice.
All the best/Evert
Here’s my workaround;it does the join using canonical SMILES, which is a bit easier than the index column and adds a column that is at least not completely useless.
Two Component Reaction.knwf (29.5 KB)
Very nice solution, thanks for the tip!