Re: [AMBER] Can amber16 pmemd rigorously combine position restraints with pressure coupling on GPUs ?

From: David A Case <>
Date: Mon, 29 Jan 2018 11:45:36 -0500

On Fri, Jan 26, 2018, Chris Neale wrote:
> I'd like to restrain part of a protein that is simulated in a box of water
> and avoid any artifacts from pressure coupling. Therefore, I'd like to
> never scale or shift the coordinates provided to pmemd via -ref and have
> the restraints at all simulation steps be independent of whether or not
> there have been restarts between those steps.
> To implement the restraints, my pmemd -i file contains:
> ntr=1,
> restraint_wt=2,
> restraintmask=':1-50.CA',
> I saw some suggestion by Tom C. that this might be possible via code
> modification here:
> "(2) Alter runmd.f to NOT call ew_pscale IF (i.e. comment out two
> lines of code)."
> but that advice is 10+ years old and amber16 seems not to have a file
> called runmd.f.

File is now called runmd.F90. I think the function you are probably looking
for [to skip it] is pressure_scale_restraint_crds().

> So, is amber16 capable of rigorously combining pressure coupling with
> position restraints on more than one atom, especially given potential
> simulation restarts? I realize that there are some workflows in which the
> problems might not be horrendous, but I'm really hoping there is a
> completely proper way to do this.
> The ability to never scale reference coordinates is probably what I'm after.

I think Tom's advice is still worth trying: of course, you will want to
carefully examine short runs to make sure you are getting what you want.

Barostat experts might want to chime in here. I'm not convinced that the
words "rigorous" or "completely proper" apply to what you are proposing,
or to anything that combines pressure with artificial forces. But what
you are proposing sound reasonable on its face.

...good luck....dac

AMBER mailing list
Received on Mon Jan 29 2018 - 09:00:02 PST
Custom Search