Feature Remover Enhancement Request

Hi Dmitry,

Is it possible to have a new feature inside the "Feature Remover" node called "Remove Heteroatom Variations". The effect of this feature is that all heteroatoms such as O, N, Cl, F, S etc are all replaced with Carbon atoms to give in effect a carbon framework of the molecules.

The reason this would be a really nice feature is that these resultant carbon skeletons or frameworks can then be used inconjunction with the Scaffold Finder node to effectively give the Most Common Skeleton or Most Common Framework. This will help the medicinal chemist to understand the minimum shape/framework requirement within a chemical series.

 

Thanks,

Simon.

Simon,

Replacing all heteroatoms with carbon will cause trouble. Valence on some of the new carbons will become incorrect (take 6-connected sulphur or 5-connected nitrogen for example). Molecules with incorrect valence can not be saved to SMILES or involved in substructure matching.

We could try to fix that by setting explicit valence ( C(V), C(VI), etc), but such thing seems hard to do when the atom is in aromatic ring. Aromatic rings with all heteroatoms replaced with C will look incorrect, by the way.

How about replacing all the atoms (carbonds and heteroatoms) with a pseudoatom that has some predefined label, for example asterisk "*"? That would cause less trouble, as pseudoatoms can not cause valence errors.

Best regards,

Dmitry

Yes that would be fine also, so long as everything is uniform in such a way as to directly compare the scaffolds with each other. I am happy with A atoms or something similar. I didnt take account of Sulphur (such as RSO2R groups), which as you say will cause trouble if it is replaced with carbon.

Thanks,

Simon.

Simon,

I have added the  "Atom Replacer" node. It will be available tomorrow in the new nightly build.

Best regards,

Dmitry

Brilliant, I am really happy that you have allowed the option to choose the atom to replace to, i.e. *, or C etc.

Thanks very much, this is working really well with my Murcko scaffold analysis. Its really helping in looking at the common scaffolds on the SAR platform and seeing how many exemplifications there are. It gives a good idea as to the minimum requirement for the biological receptor in terms of a pharmacophore.

Simon.