[gmx-users] Additional repulsive terms for LJ

[Justin Lemkul]

On 8/25/14, 3:35 AM, Timothy Click wrote:
> I am simulating a system in N,N-dimethylacetamide (DMA) and ions (Li+ and
> Cl-) using CHARMM36. However, I have found that the ions tend to cluster
> together because of the lower dielectric constant of DMA.If I lower the
> formal charges (+/-0.5), the clustering diminishes considerably. A former
> postdoc in the group had used NAMD when simulating a similar system and
> simply added an additional repulsive term to the Li-Cl LJ interactions to
> remove the clustering. Before I attempt anything, I wanted to ask if this is
> possible in Gromacs by adjusting parameters in the [pairwise] section of the
> ffnonbonded.itp. Or, is the adjustment of the formal charge an acceptable
> physical representation because of the difference in solvents (water for the
> CHARMM36 parameters of LiCl vs. DMA with LiCl)?
>

Adjusting the charges is unphysical, and affects all interactions, thus 
undermining the whole force field.  The easiest approach is to add a suitable 
entry in nbfix.itp to set special [nonbond_params].  This is often used in 
CHARMM (the "NBFIX" function) to set special interactions for ions, since 
additive models of ions are often very crude.  Obtaining proper parameters is 
very laborious and should not be simply set ad hoc to make something go away. 
The way we often tune these interactions is by doing osmotic pressure 
calculations, though we do them in water, so other solvents should be approached 
with caution, anyway.

-Justin

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

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

Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 601
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

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

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