[gmx-users] Minimum distance periodic images, protein simulation

[Justin Lemkul]

On 9/20/13 4:11 PM, Arun Sharma wrote:
> Hello,
> I ran a 100-ns long simulation of a small protein (trp-cage) at an elevated temperature. I analysed the distance between periodic images using
>
> g_mindist -f md-run-1-noPBC.xtc -s md-run-1.tpr -n index.ndx -od mindist.xvg -pi
>
> The output shows that there are situations when the closest distance between certain atoms is much lesser than 1 nm. Conventional wisdom says that if this happens the simulation results are questionable. Is this completely true? If this is indeed true, how would I ensure that this does not happen again?
>

Using a sufficiently large box (minimum solute-box distance at least equal to 
the longest cutoff) is the general procedure.

> I have posted the output of g_mindist at http://postimg.org/image/bnc0ej3nb/
>

Some configurations definitely come very close, indicating several frames with 
spurious forces throughout the duration of the trajectory.  I would be very 
suspicious of the results.

-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

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