I've been using the substructure match with query molecules to do an R-Group decomposition. This works well, except in the case where the query molecule contains a 'bridged atom in a ring. Take this example (I've replaced the irrelevant part with a Y atom for confidentiality)
FC1=CC=C(N2CC3CC2CN3S(=O)(=O)C2=CC=C([Y])C=C2)C2=CC=CC=C12 [image: image.png] Using the following as a query, you get the usual result you'd expect: *-N1CC2CC1CN2S(-*)(=O)=O [image: image.png] However, I also see when I do the match: [image: image.png] Even switching back to the R-group code from before the latest refactor seems to have this issue, like the query molecule is perceived as having the actual bridged ring, and the smaller ring bounded by the bridge atom. Am I missing an obvious setting to exclude those latter matches, as obviously the groups generated do not match reality in this case. Such as it is, I'm doing everything via a version of Pat Walters R-Group method from the older RDkit, with the list to generate SMILES coming from: match_list = test_mol.GetSubstructMatches(self.query_mol, False)
_______________________________________________ Rdkit-discuss mailing list Rdkit-discuss@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/rdkit-discuss