I've been looking into Free-Wilson, attempting to rework the Pat Walters code (https://github.com/PatWalters/Free-Wilson) for doing this to something more suited to batch use.
With the RDKit having a full RGroupDecomposition library that can generate the data directly, this seems like a good move, but my concern is about those molecules for which an R Group is linking two sections of the core, which is fairly common in Markush structures. Is this something the MCS code here will cope with, or will the fact that the R Group has multiple connections within a molecule cause problems when it comes to the decomposition? I've not seen many examples of the linker case aside from in the likes of Stardrop so it'd be nice to know if this does confer and advantage over, say, getting the substruct matches and doing the MCS and subtraction separately.
_______________________________________________ Rdkit-discuss mailing list Rdkit-discuss@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/rdkit-discuss