Dana

Is there any way to use a chiral fingerprint, say ECFP6#S, to build models using the Learn Molecular Categories component? I've got examples where the biological activity is highly dependant on the chirality, and the models should try and capture that.

(the same goes for the Random Forest components)

By the way, the model evaluation needs to take chirality into account. If the model says that the existence of a particular "up" methyl contributes to some activity, if you give it an achiral methyl (i.e. the racemic mixture) it should still consider the contribution of the chiral methyl in the model even though the racemate is only one half "up".

Lee