[gmx-users] Separate permeating water molecule from solvent water

[Justin Lemkul]

On 2/28/16 8:42 PM, Michail Palaiokostas Avramidis wrote:
> Dear GMX users,
>
>
> I am trying to simulate water permeating through a lipid bilayer with
> CHARMM36. I am doing this by using the pull - constraint algorithm to
> constrain the distance of water and membrane to a specific depth. I have
> noticed that the simulation crashes with LINCS warnings at a random time,
> even after 10s of nanoseconds, only when I simulate water and not other small
> molecules.
>
>
> I have read in the GROMACS documentation that if a water molecule is
> separated from the rest of the solvent, this might create instabilities. So
> far I have been treating the water permeant as part of the solvent in the
> topology file and therefore, I am convninced that the mistake is there.
>

I disagree.  Topological separation and physical separation are different things 
entirely.  Separating a water molecule topologically will have no implication on 
the physics because the interaction parameters are all identical, you've just 
assigned a new name to one molecule.  Likely what you have seen are past posts 
that refer to water molecules that get trapped in small vacuum bubbles in some 
systems; those can become destabilized because they tend to crash around in the 
vacuum and disrupt nearby molecules.  I see no reason why a water in a membrane 
would be unstable (people have observed these in unbiased MD simulations with 
GROMACS), and certainly playing topology tricks won't change that.

Likely whatever your protocol is "works" well enough for other molecules, 
perhaps because they diffuse more slowly or interact more favorably with the 
membrane, and water presents a pathological case.  You should re-evaluate your 
methods.  The constraint mode for pulling can be somewhat unkind, as the 
constraint distance must remain fixed over time and water may diffuse quite 
rapidly.  Perhaps a shorter time step will help in keeping the constraint happy. 
  Otherwise, explore other pulling options.

>
> I would therefore like to ask how I can create a new water topology of CHARMM
> TIP3P just for the permeating water.
>
> Shall I just copy the tip3p topology that is included in CHARMM36 folder in a
> new itp?
>

As stated above, this is highly unlikely to make any difference.  You can create 
a duplicate water [moleculetype] with a different name, but you'll also have to 
replace the [settles] directive with three normal [constraints].  Troubleshoot 
your existing data before going this route.

-Justin

-- 
==================================================

Justin A. Lemkul, Ph.D.
Ruth L. Kirschstein NRSA Postdoctoral Fellow

Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 629
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

jalemkul at outerbanks.umaryland.edu | (410) 706-7441
http://mackerell.umaryland.edu/~jalemkul

==================================================