[gmx-users] REMD general protocol ...
jalemkul at vt.edu
Tue Apr 2 18:34:45 CEST 2013
On 4/2/13 7:38 AM, Erik Marklund wrote:
> On 2 Apr 2013, at 13:30, Justin Lemkul <jalemkul at vt.edu> wrote:
>> On 4/2/13 7:13 AM, rama david wrote:
>>> Dear friends,
>>> I am naive to the Replica exchange Molecular dynamics ( REMD).
>>> I have plan to use REMD for temp. 310-320 K to my system.
>>> I thoroughly search the Mailing-list Archive for the REMD problem.
>>> It was a really helpful to start.
>>> My system consist of peptide + water.
>>> I used the following work-flow, Would you please help me to find out
>>> my mistakes...
>>> 1. energy minimesation for peptide + solvent
>>> 2. Different Mdp file for temp. ( 4 temp therefore 4 mdp file )
>>> 3. Make tpr file for each nvt run
>>> 4. Then separate equilibration for each temp ( 4 equilibration steps )
>>> 5. Then made NPT.mdp file for each temp ( 4 temp )
>>> 6. Then again equilibration for NPT at 4 temp.( 4 equilibration steps )
>>> 7. Then run md production with -replex 1000 -multi 4 command ..
>>> To determine the temp I used web-server http://folding.bmc.uu.se/remd/
>>> Please suggest me any improvements that are possible to implement in
>>> my work flow.
>> Such a narrow range of temperatures defeats the purpose of using REMD. Normally, a much larger range is used over many more simulations. For near-ambient temperatures, NPT can be used, but if you include much higher temperatures, you should use NVT due to box instability upon exchanges.
> Sure, the enhanced sampling is basically gone, but you can deduce temperature dependences from such simulations and to some extent benefit from the mixing, can't you?
I suppose, but then why bother with the exchange overhead? Seems to me that if
you're only interested in temperature-dependent quantities, you can do that with
independent simulations at different temperatures.
Justin A. Lemkul, Ph.D.
Department of Biochemistry
jalemkul[at]vt.edu | (540) 231-9080
More information about the gromacs.org_gmx-users