[gmx-users] VDW Cut-off vs Switch with CHARMM force field

[Justin Lemkul]

On 8/24/13 11:11 PM, Gianluca Interlandi wrote:
>> vdwtype = switch
>> rlist = 1.2
>> rlistlong = 1.4
>> rvdw = 1.2
>> rvdw-switch = 1.0
>> rcoulomb = 1.2
>
> I'm confused about the difference between rlist and rlistlong. I am planning to
> also setup an implicit solvent simulation with GBSA and CHARMM-FF. Would this
> make sense?
>
> rlist                    = 1.4
>
> coulombtype              = Shift
> rcoulomb_switch          = 0
> rcoulomb                 = 1.2
>
> vdw-type                 = Switch
> rvdw_switch              = 1.0
> rvdw                     = 1.2
>

Implicit solvent simulations are a special case, and the cutoffs used are 
generally different than in explicit solvent.  In fact, I would be willing to 
bet that the settings above would lead to unstable structures.  Note that you 
should not use shifted electrostatics, as they produce significant artifacts, 
and the fact that you'd be shifting over the entire range (0 to 1.2 nm) is a 
very bad idea, indeed.

With implicit solvent, literature often refers to longer cutoffs, on the order 
of 2.0 nm.  In my hands, such simulations have never worked.  Perfectly stable 
proteins unfold and energy drifts badly.  The only stable way I have found to 
run GBSA simulations is using the all-vs-all approach:

pbc = no
ns_type = simple
rvdw = 0
rcoulomb = 0
rlist = 0
nstlist = 0

Note that the use of zero cutoffs does not mean that interactions are not 
calculated; just the opposite - all interactions are calculated.

-Justin

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

Justin A. Lemkul, Ph.D.
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

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