Molecule fragmenter


i have had a few attempts with the fragmenter node but so far have not managed to find it very applicable to real life problems. I often find the fragments are too far removed from the original molecule. Maybe I am mis-using it. It just generates too many fragments.

is it possible to put a restriction on the fragmenter to not break up rings. I.e. do something like the waldmann scaffold tree fragmentation approach where it sequentially breaks off the least significant ring system, or it generates all possible fragments of breakages next to ring systems.




Hi Simon,

The fragmenter is really designed to provide input for machine-learning or statistical methods. 

To have things that look more like real molecules or at least recognizeable pieces thereof, it seems like something like a "retrosynthesis" node that does either RECORE or BRICS analysis would be useful.

Sound right?


Hi greg,

A retrosynthesis node using RECORE/BRICS analysis to generate fragments using chemistry rules would be great. Purpose would be to use these fragments in a number of ways, such as in scaffold replacement tools and reaction enumeration nodes, or to use them to cluster sets of molecules together.