Reply To: modeling residues with covalent ligand attachments

[Anonymous]

I’m writing my reply for the “general release” audience, Sid – but why not just put this on the mini mailing list?

I’ve done something similar. In one case, I was modeling a covalently attached ubiquitin/E2 system; the paper just came out:

http://www.ncbi.nlm.nih.gov/pubmed/21474069
Mol Cell. 2011 Apr 8;42(1):75-83.
Essential role for ubiquitin-ubiquitin-conjugating enzyme interaction in ubiquitin discharge from cdc34 to substrate.
Saha A, Lewis S, Kleiger G, Kuhlman B, Deshaies RJ.

This code will release with 3.3 (which is planned for approximately August.) In this work, we used a molecular-mechanics derived potential (within Rosetta) to sample the thioester torsions directly, then scored the docking of the two proteins with standard Rosetta score12.

I’ve also modeled a covalently attached dye (post-translationally added to a surface cysteine). This paper isn’t out yet, but the paper doesn’t describe that aspect of the code very much anyway; that code will also release in 3.3. In this work, we treated the whole sidechain as a noncanonical amino acid, generated a rotamer library for it, and just let Rosetta treat it normally in packing.

You will be interested in this other thread about generating noncanonical AAs (another cysteine-linked experimental reagent); you may want to grab the tarball there while it’s up: http://www.rosettacommons.org/content/some-application-troubles-when-using-resfile-flag#comment-2626

Some of the code for doing these things hasn’t been released and probably won’t be for a while. Also, beyond the Meiler and Kuhlman labs, you may want Doug Renfrew (the NCAA guy) on your radar, as he is the best at getting your chemically-conjugated inhibitor into Rosetta as a valid residue type.