Re: [AMBER] net charge issue during constant pH simulation with explicit solvent model

From: Jason Swails <>
Date: Wed, 18 Nov 2015 11:40:32 -0500

On Wed, Nov 18, 2015 at 2:19 AM, zhonghe xu <> wrote:

> Hi Jason:
> I appreciate your reply about the temperature issue. I have other concern
> about PME calculation in
> constant pH simulation with explicit solvent model. Due to the net charge
> of system, Ewald or PME
> calculation may be generate some error.

​The standard PME implementation uses a net neutralizing plasma that
basically smears a neutralizing charge distribution evenly across the
grid. There are papers out there that quantify this effect, and for all
but small unit cells the effect is fairly small (certainly smaller than
many *other* sources of error). The biggest artifacts have to do with
computing free energies and pressure equilibration. Since the approach
here uses GB to do the protonation state changes, the FE artifacts don't
affect the titrations at all, and the pressure artifacts are small for
standard unit cell sizes (and don't affect NVT simulations which are
required for pH-REMD at the moment).

> In order to circumvent this issue, is there other method in Amber 14
> to handle long coulomb calculation for system running with periodic
> boundary condition?

​No. But the errors/approximations associated with those other methods
(like reaction field or IPS) don't seem to me to be any less severe (and
are perhaps even more severe) than the effect of the neutralizing plasma.


Jason M. Swails
Rutgers University
Postdoctoral Researcher
AMBER mailing list
Received on Wed Nov 18 2015 - 09:00:04 PST
Custom Search