Use dihedral angles to distinguish between different stereochemistry

[bieniekmateusz]

Long term goal to sort out when we have a minute. Let's stay in touch about this!

[shunzhou]

I agree we need to address it. I think we should first identify stereocentres for each of the compounds, and then compare them to see if they are identical if the stereocentres appear in both compounds. If they only appear in one of the compounds, we still need to use the stereocentre information to correctly construct the structures of the hybrid compounds.

Hydrogen atoms can't be hidden anymore.

[shunzhou]

It's better to get this sorted before we release the code.

I don't think normal dihedral angles are of any use here.

Improper dihedral may be useful, as one of the purposes of introducing improper dihedral is to prevent atoms from flipping over to their mirror images.

Is there anything we can learn from the identification of tetrahedral atomic stereochemistry in other packages such as RDKit?

[bieniekmateusz]

It would be useful to have it for the release. I'll use rdkit to add this case. Let's mention this on the next meeting. I will try to take care of it by then.

Best, Mat

[bieniekmateusz]

A quick update, I used rdkit to detect the enantiomers we previously encountered and they didn't manage it in that one case. Specifically, I used this:
https://www.rdkit.org/docs/source/rdkit.Chem.EnumerateStereoisomers.html
on these two molecules
mols.zip
this case might be one of the harder ones but it might be worthwhile to consider it from the point of view of the forcefield. Ie compare the applied dihedral / angle terms and exclude them if they're different, which they should be. I'll investigate that path next.

[shunzhou]

The force field terms are very likely to be exactly the same for the two molecules. We may need structural information instead of force field terms to detect the differences of the conformation.

[bieniekmateusz]

In that case we might need a better set of molecules and clearer definition. I am happy to use rdkit where we can, but it might be also good to check the libraries.

I am surprised about the terms in the FF. But I can see how they could be in different minimas while having the same FF.