[gmx-users] Calculating avergae structure from the simulation

[Justin Lemkul]

On 6/8/16 5:21 PM, Agnivo Gosai wrote:
> Dear Users
>
> I had run an unrestrained 50 ns simulation of a protein-ligand complex
> using GROMACS to check the stability of the force field used and compare
> with previous literaure.
>
> Then I dumped the strucuture from the last frame of the 50 ns simulation
> and used it for pulling and umbrella sampling simulations.
>
> Now, the reviewer is asking me this "Reference structure should have been
> the average structure, not the final state seen at the end of 50 ns MD run.
> "
>
> Is there any tool in GROMACS by which I can calculate the average structure
> ?? I was thinking that this is a normal procedure to conduct the
> simulations.But now I am confused. Kindly comment.
>

The final structure may or may not be representative of the simulation.  What if 
your protein twitched in some way in that last frame that is totally unique to 
the trajectory?  Maybe it is representative, maybe it's not.

You can calculate an average structure, but it may not be physically meaningful 
(see http://www.gromacs.org/Documentation/Terminology/Average_Structure), but 
you'd be better off clustering the conformations, determining the largest 
cluster, and if that cluster is sufficiently representative of the overall 
ensemble, use the central member (lowest RMSD with respect to the other cluster 
members) as your representative snapshot.

-Justin

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

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

Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 629
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

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

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