[gmx-users] Why should I repeat my simulation?

[Justin Lemkul]

On 10/4/14 5:15 PM, fatemeh ramezani wrote:
> Dear gmx-users
>
> I performed a MD simulation with gromacs software and after sending my results to a journal, referee said
>
> "Simulations needs to
> be repeated with different initial velocity (seed) to make sure that the
> results are not simulation artifacts.It is needed to ensure that the simulation represents true dynamics
> of the system."
> It is really need to I repeat my simulation (that is very time-consuming process) or you can help me to providea compelling reason for not repeating.
>

The reviewer is right.  A single simulation is generally insufficient to provide 
convincing evidence of any behavior, unless it is exceptionally long and shows, 
for instance, reversible behavior.  In reality, no simulation is probably ever 
truly converged.  That is why it is standard practice to conduct multiple 
simulations, exactly as the reviewer says.  One would never perform a single 
biological assay and claim it to be exactly correct, why should one simulation 
be the exact answer?  Perhaps your one simulation would end up being the outlier 
in a data set.

-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

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