Is there a node or strategy I can use to identify any constitutional isomers present in a large chemical dataset?

KNIME Analytics Platform
1

I have a dataset of >2000 compounds. I am looking for a quick method to see if there are any constitutional isomers present. I am thinking maybe I need to generate fingerprints and look at similarities?

2

Hey Tova - Welcome!
I don’t have a solution for this, but could you use the OMG.jar library via the External Tool node to generate all the isomers for each of the 2000 compounds, one compound at a time in a loop, then search the remaining compounds for matches? If you used the Recursive Loop nodes, you could eliminate all matches in one loop iteration from the need to do a subsequent search on the next iteration.

Or perhaps the more recent MAYGEN.jar library? The OMG.jar didn’t work for me with java21.

There’s also the RetroPath paper with associated KNIME workflows:

(the other)
Simon