Re: [AMBER] pH-REMD theory

From: Jason Swails <>
Date: Thu, 13 Aug 2015 08:43:41 -0400

On Thu, 2015-08-13 at 14:58 +0800, whb wrote:
> Hi all:
> I have a question about the method of pH-REMD. I’ve read the paper
> published in JCTC(J. Chem. Theory Comput., 2014, 10 (3), pp 1341–1352).
> It said that the probability of accepting these replica exchange attempts,
> given by eq 3, depends only on the difference in the number of titrating
> protons present in each replica and their respective difference in pH
> I think that it contains an assumption that the residue in lower pH can be
> more protonated. But I think that pka of the residue depends the environment
> where the residue exists in the protein. And when the proteins structure has
> changed severely in lower pH , the residue may have another pH in that
> situation. And this situation may contradict with the assumption the theory
> contains.

There are indeed a lot of assumptions made in any MD simulation --
pH-REMD included. However, the *replica* exchange acceptance
probability that you refer to is derived directly by solving the
detailed balance equation of the probability distribution functions for
the semi-grand canonical ensemble. Therefore, this is one of the few
aspects of the simulation that is, in fact, *not* an approximation and
is exact.

The effect of the protein environment is accounted for by the
protonation state change attempts *within* a replica (keep in mind there
are two Monte Carlo change attempts being made -- one between
protonation states and the other between replicas). I present an
abbreviated derivation of this equation on my Wiki: and there are also more
detailed explanations about replica exchange in general as well as the
basic theory underlying constant pH simulations in my dissertation:


Jason M. Swails
Rutgers University
Postdoctoral Researcher
AMBER mailing list
Received on Thu Aug 13 2015 - 06:00:03 PDT
Custom Search