You are here

A question on the logic of enzyme design

3 posts / 0 new
Last post
A question on the logic of enzyme design

I am trying to use enzyme design app in Rosetta 3.4 with a small ligand docked into an enzyme (no enzyme-ligand covalent connection), and I have two questions about it.

The first is, I do not know which residues of the protein take part in the catalysis, so I cannot form a .cst file. Would results of such an enzyme design job without a .cst file be somewhat reliable or total rubbish? What else could you suggest me to do about it, is there any way to compensate the absence of such catalytic knowledge, or is the program only designed to work with appropriate catalytic constraints?

My second question is, at the moment, I made enzyme design application work. There is no .cst file as input, only the flags, unbound pdb, and enzyme docked ligand pdb. The resulting job finishes gracefully and gives some final structures, all being identical (also with identical scores). Why could this be happening? The flag file I used is what is suggested by the demos, with only minor changes.

I would be glad if you could help me with these,

Post Situation: 
Thu, 2013-06-13 08:10

If you don't need to specify protein-substrate geometry, you don't need to supply a cst file. It should work perfectly fine without. In fact, that's how we benchmark the protocol - we run the protocol against ligand binding proteins without any constraint files.

The limitations of not using the cst file is that you can't encode the additional information about the catalytic geometry that Rosetta (not being QM) won't know about. So Rosetta might do things like move acid and base groups from where you want them to be, or even mutate them to a completely different group. The latter can be addressed by giving the protocol a resfile, but for geometry issues, you'll need to use a cst file. (Though you can try without first, and then add a cst file later if you decide to.)

On the second question, one possibility is that your system is highly convergent, and Rosetta can find the best structure (to its mind) every single time. I think this is unlikely, though. More likely is that something in your flags file is unduly limiting the design procedure. I'm not quite sure which demos you're following. Could you post the commandline and flags file you're using?

Also, take a look at your starting structure. The usual enzyme design parameters repack and design in a region around the ligand. If your ligand isn't contacting the protein, you might not actually be considering any changes. In the logging output for the run there should be an indication as to which residues were included in the repack and design shells. See if the number and identity of those residues match up with what you expect.

Thu, 2013-06-13 11:11

Thanks a lot for the help. It appears that I only missed adding cst_design keyword (thinking that I would not need it since it contains a cst command) to switch on the design. As a result, I got no results from the design part. Reading the flags more carefully, I added it back. After adding it, log file started to mention non-zero design shells and hopefully will be okay. Also thanks for the view on working jobs without a .cst file.


Mon, 2013-06-17 11:36