Possibility of adding non-canonical amino acids

[amin-sagar]

Hello.
Thanks for this awesome work.
I would like to run side chain optimization routines for a peptide with non-canonical amino acids. Is there a way to add such residues in the rotamer library?
I would be really grateful for any suggestions.
Best,
Amin.

[mallory-tollefson]

Hi @amin-sagar. Thanks for your interest in our work! The simplest approach to optimizing a peptide with non-canonical amino acids would be to temporarily convert those residues to their closest analog among the canonical 20 amino acids, exclude those amino acids from the list of residues to be optimized, and replace the analogs with the non-canonical amino acids after the optimization is complete. This would generate a protein structure that is optimized apart from the non-canonical amino acids.

Adding non-canonical amino acids to this optimization protocol is an interesting application, particularly for protein design or studies of species with different native amino acids, but unfortunately, it isn't a quick addition to the algorithm at this time. To truly optimize non-canonical amino acids with this many-body algorithm, one would need to 1) generate a set of "rotamers" for each non-canonical amino acid based off the statistically prevalent conformations for those amino acid side-chain atoms seen in experimental structures, and 2) generate force field parameters for each non-canonical amino acid with software such as Poltype and add these new parameters to the force field.

I hope the "quick fix" option works for your applications, and thanks for sparking ideas about formally adding non-canonical amino acids in the future!

[amin-sagar]

Thanks @mallory-tollefson

The approach of converting the non-canonical amino acids to closest natural amino acids would work for some of my cases where there is a small change like methylation or halogenation etc.

However, I would also like to work with more radical changes. I have rotamer libraries of some of these residues generated in MOE using the LowMode approach (https://pubs.acs.org/doi/10.1021/ci900508k). Additionally, I can try to use the approach described in https://pubs.acs.org/doi/10.1021/acs.jctc.0c00648.

Could you point me towards some documentation about the format of the rotamer library that forcefieldx accepts?

Best,
Amin.