Having generated conformers from some vendor libraries, I've noticed the presence of cis-amide bonds (i.e. conformer sets for a compound are a mix of cis and trans) that I'd like to avoid. Is there any way of not generating these conformers or filtering them out ?




At the moment the conformer generation node doesn't provide sufficient control to allow you to avoid this problem. The new RDKit conformer generator (from last year) does have an option to lower the likelihood that you'll get one of these (by using experimental torsion preferences), but you cannot (yet) turn that option on from KNIME.

We'll try to enable that relatively soon.



Thanks Greg & sorry for a very late reply.

Any idea if one can somehow catch/filter the unwanted conformers after they're already generated ? I noticed the problem too late and carried the conformers into a relatively lengthy docking run and would prefer to filter out the wrong poses rather than repeating the whole run.



PS. Having experimented a bit, I did notice that the cis-amides don't show up with KNIME when running the conformer generation with implicit hydrogens. Is this a valid approach or should one always add explicit ones ?

You can certainly do it using the RDKit code in the Python node (assuming you have an RDKit install for Python), but I'm not aware of any other way to do it.

And to the PS: It's almost always a better idea to add explicit Hs to your molecule before generating conformers.




Could I check - is this functionality available in the stable RDKit builds yet, or still only the nightly?  If the latter...?



Hi Steve,

Still only the nightly builds. We'll move it over to the stable builds once we update the binaries to reflect the most recent RDKit release. This requires that the release actually be finalized, but that should be taken care of this week.

In the ideal case, we'll have a production version of the new nodes sometime next week.



Thanks Greg, that's very useful to know.