On Tue, Apr 7, 2015 at 2:33 PM, Emilio Angelina <
emilioluisangelina.hotmail.com> wrote:
> Hi,
>
> I would like to compute some properties derived from charge density on one
> or more snapshots from molecular dynamics trajectories by performing QM
> calculations.
> Since computing these properties on a single structure might be
> meaningless, I was thinking in performing a clustering analysis using
> ptraj, pick a representative structure from each cluster and compute the QM
> properties on each of those representative structures.
> Then, my problem is how to combine the QM results from all the snapshots
> to get a single result, for example a single interaction energy that then I
> can compare with binding affinity experimental data.
> Would it be possible to weight the results from each representative
> structure by its Boltzman weight?
> Or should I just average the results?
> I know that the best way of taking into account the Boltzman distribution
> would be to take a very large number of independent snapshots from
> trajectory just like in MM-PBSA but, since the QM calulations are too
> expensive I only can take a small number of frames, so I would need a way
> to explicitly compute the Boltzman weights.
>
I would do a weighted average by the cluster populations (which should,
themselves, be Boltzmann-weighted by virtue of running a MD or proper MC
simulation).
So the average I would do would be something like this:
<A> = SUM_x( Nx * A(X) ) / Ntot
where the summation runs over all clusters (x), X is the centroid of
cluster x, Nx is the number of points in cluster x, and Ntot is the total
number of points in the trajectory.
HTH,
Jason
--
Jason M. Swails
BioMaPS,
Rutgers University
Postdoctoral Researcher
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Apr 07 2015 - 12:00:03 PDT
