You are here

Enzyme Design with additional covalent backbone bonds

2 posts / 0 new
Last post
Enzyme Design with additional covalent backbone bonds

Hey everyone,

I am trying to design a protein with a covalent bond between the C of Thr and the N of Gly with a Tyr in between them (imitating the chromophore from GFP for perspective). Because of this odd 5-membered ring formation, I used Thr/Gly with a generous distance constraint between them in my CST file when I ran RosettaMatch to find likely sites in my scaffold proteins. My sites place the two backbone atoms ~4-5A apart.

However, now I need to run enzyme_design on my best sites and get the correct formation of this backbone bond. This seemed like the right time to experiment with patches for Thr/Gly to allow the extra backbone bond because the final chromophore structure just removes the O on Thr and the H on Gly. I've read a bunch of forum posts about patches for modifications and looked at most of the patch files. I mostly based mine on the side-chain modification patches in Branching/Conjugation folder. Using a patch should also mean that centroids are handled easily. 


1. If I use Thr/Gly as residues names in my input PDB (as pointed out as being correct in several forum posts on non-canonical AA) with a link record and use T/G in my CST, Rosetta doesn't seem to recognize that I really mean a patched Thr/Gly residue based on the atoms present in the residue. I do get a warning about missing heavy atom O on Thr. Enz_design runs to completion and gives me a structure with perfectly reasonable looking score components and obviously no backbone interaction. 

2. I use TRX/GLC (the modified names in my patch files) in all relevant input files, adding a LINK record to the PDB. This fails, with this error:

protocols.toolbox.match_enzdes_util.EnzConstraintIO: read enzyme constraints from 2y0gAAbindingHOHMod.cst ...

ERROR: Failed to apply a PatchOperation to THR

ERROR:: Exit from: src/core/chemical/ line: 333

Obviously, something is wrong with my patch, which probably explains why I don't seem to get the additional backbone bond in #1. I added my patches to the patches.txt file. 

3. CRO is an existing params file in fa_standar/residue_types/extra that codes the whole 3-residue chromophore from GFP. I tried replacing the relevant three residues with CRO and adjusting atom names, etc. But there's no corresponding centroid, which would cause problems later. And it also fails with the same "Failed to apply a PatchOperation" error as #2. 

Any ideas? Do I need to reevaluate and go all the way back to RosettaMatch? Or just fix my patches? Do I need to switch to using RosettaRemodel, where I'd probably have all the same patch issues? I'm attaching both patches and my PDB file. The relevant residues are T23 and G25.  Thanks!


THR patch464 bytes
GLY patch464 bytes
1QD5_HOH_9_Mod.pdb320.31 KB
Post Situation: 
Wed, 2016-08-03 10:23

Have you tried using a covalent constraint? If you have an enzdes-style constraint file, you can specify a "covalent" constraint by setting the "periodicity" column of the distanceAB line to "1" instead of "0". This will cause Rosetta to turn off the repulsive term between these atoms and surrounding atoms. I haven't tried it with constraints between two polymeric residues, but I don't see a priori why it shouldn't work with standard amino acid residue types.

Mon, 2016-09-05 11:29