[gmx-users] Gas phase simulation

Christopher Neale chris.neale at alum.utoronto.ca
Tue May 5 14:33:02 CEST 2015

Dear Gozde:

There are a few questions:

1. Does your system remain triphasic? You can check quickly one system with VMD or you can do it quantitatively with g_density (separate profiles for all-atoms, water, organic solvent). I see a triphasic system in your image but that doesn't guarantee that it stays this way.

2. I'm presuming that you added the restraints on the organic molecules because they were boiling away. Why is this? I don't know. Is your organic solvent expected to be volatile at your simulation temperature? Maybe the ff is wrong or maybe this is what should be happening.

3. I can't address how to correct for the presence of restraints unless you tell us how did you add these restraints? If you're doing absolute position restraints then I'd say you're doing it wrong. Here's what I would do.

a) simulate a biphasic water/organic solvent system (no vacuum) at NPT.

b) find the average box volume from that simulation

c) select a frame that is closes to this average volume

d) extend your box along z to create a vacuum region

e) add an enormous number of flat-bottom position restraints (one for each molecule) along z to keep the water in the water region and the organics in the organic region but to not apply any force while they are in that region.

f) do your umbrella sampling on this system using NVT

*** You're not going to get the interfaces correct with this design. Therefore, your profiles will be wrong but the overall dG of transfer should still be correct. You could probably use only a single flat-bottom half-harmonic on each organic solvent molecule to stop it from evaporating and let the water/air and water/organic interfaces sort themselves out. That way, you only really perturb the organic/air interface (and then there is also no need for the initial NPT to evaluate average volume and also you can do the US in NPT).

Nevertheless, the above is just one hack to force your system to behave as you are asking it to. It's probably better to figure out i) if it is behaving physically realistically and ii) if not then why not (i.e., adjust the parameters).


From: gozde ergin <gozdeeergin at gmail.com>
Sent: 05 May 2015 07:16
To: Christopher Neale
Subject: Fwd: About 2011 paper

---------- Forwarded message ----------
From: gozde ergin <gozdeeergin at gmail.com<mailto:gozdeeergin at gmail.com>>
Date: Tue, May 5, 2015 at 12:13 PM
Subject: About 2011 paper
To: nealec at rpi.edu<mailto:nealec at rpi.edu>

Dear Dr. Neale,

After I read your 2011 paper about the umbrella sampling I would like to ask you some questions.

I am also running umbrella sampling simulation by Gromacs.
My system has 860 water molecules covered by 50 organic molecules.
My purpose is to estimate the PMF profile of a water molecule from bulk water to gas phase (reaction coordinate in order : bulk water->organic layer->gas phase) Please check 1_system.png.

In order to do that I picked a water molecule from bulk water and pulled it through the reaction coordinate. I would like see change in PMF from bulk to organic and organic to gas phase. However I could not see any difference between organic to gas phase. Please check 2_without_restraint.jpg

Thereafter I thought I maybe should restraint my organic molecules position to reduce their effect on gas phase free energy.
By doing so I managed to see the change in PMF from organic layer to gas phase. Please check 3_with_restraint.jpg

But by restraining organic molecules positions I added an extra force on system.
Now I could not figure out what should I do to minimize the effect of positions restraint?
Also another question why I could not sample the gas phase without restraining my organic molecules?

In figures, z=0 the middle of the water bulk
               z>4 gas phase
               2<z<4 organic phase

Any suggestions will be appreciated. Thanks in advance.

More information about the gromacs.org_gmx-users mailing list