Re: [AMBER] MMPBSA/Disulfide bond error

From: Jason Swails <jason.swails.gmail.com>
Date: Tue, 15 Jan 2013 17:37:19 -0500

Hi Divi,

I'm posting this back on the Amber list so you can benefit from others'
comments or arguments on my thoughts and ideas, and so that others may
benefit from the discussion in the future.

On Tue, Jan 15, 2013 at 5:13 PM, Divi <dvenkatlu.gmail.com> wrote:

> Thanks Jason. I did not think of this issue when we designed this
> project. The project was to study the mutational analysis of 10 residues at
> the dimer interface. MMPBSA energies are one key data we need to assess
> energies. Each mutant complex was simulated in water for 100 ns. All of
> them have disulfide bridge since that is essential for stability of the
> dimer.
>
> Now I guess I need to think about alternate way to get the work done.
> How about this?
> 1. I will redo all the work by removing the disulfide bridge in all
> simulations (essentially new simulations with new topology/coordinate files
> starting from the 100ns of each equilibrated complex with a layer of
> equilibrated waters) but instead of a hard-bond between the two CYX
> residues, I will impose a distance constraint of 2.5 Angs. so that the
> dimer complex does not diverge.
>
> 2. Run each complex for 10 ns (takes 5 hours at the most on 256 cores,
> about 40K atoms) fresh from start and do MMPBSA analysis again.
>

I think you are going about this the wrong way. You will never be able to
justify eliminating disulfide bonds when your paper goes to peer review.
 The warning printed in the MM/PBSA output file is just that -- it is a
warning. While in most cases it means that the topology files have been
messed up in some way, in this case it indicates that a bond is missing in
the unbound states. Even if you could impose the 'correct' distance
restraints to mimic a disulfide bond (which you can't, due to the way Amber
handles 1-2, 1-3, and 1-4 nonbonded interactions), all you've succeeded in
doing by this approach is tricking MMPBSA.py into not realizing there's a
disulfide present so you can eliminate the warning message.

What you should do instead, in my opinion, is to think of how you can use
the data from MMPBSA.py (and what that data really means) in your arguments
and analyses. You know that there are effects involved with using MM/PBSA
with a covalently bound complex -- the bond breaking energy from the force
field is fake (and wrong with respect to experiment), there's going to be a
difference in charge distribution associated with that broken bond, the two
cysteines may be reduced, or they may stay in their negatively-charged
state (you need to determine which is the case), and if your cysteines are
reduced, you have introduced more particles in your products that you must
account for in your reactants, etc.

Depending on how you are comparing your mutants, though, you may be able to
make the argument that these effects will be the same in both forms of your
system, so its effect in the binding energy difference cancels. It is
equally important to look at reasons why this approximation may fail, and
present counterarguments whenever possible to defend what you've done. A
particularly important consideration is how close the mutation is to the
broken disulfide -- the closer it is the more likely your approximations
break down.

HTH,
Jason

-- 
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
352-392-4032
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Jan 15 2013 - 15:00:03 PST
Custom Search