Hi,
This is by no means the only way, but we've done this kind of analysis
recently (see
http://pubs.acs.org/doi/abs/10.1021/jp4125099). The idea
is to calculate principal components from a combined trajectory (e.g.
trajectories T1 + T2 + T3), then calculate projections along those
principal components separately (i.e. the projections for T1, T2, and
T3 along the combined PCs). The distributions for the individual
projections for PCs (particularly the low frequency ones) should
overlap pretty well, otherwise the trajectories are not converged -
note that overlap is a necessary but not sufficient condition of
convergence, ideally you will look at several different properties to
ascertain overall convergence. We used Kullback-Leibler divergence to
quantify the overlap of distributions, but that isn't the only way.
The SI in the given publication has some example scripts for CPPTRAJ
you should be able to adapt for your use.
Hope this helps,
-Dan
On Wed, Aug 5, 2015 at 2:24 AM, George Tzotzos <gtzotzos.me.com> wrote:
> I’ve run 5 independent MD simulations of the same system. I’m seeking advice on how to compare the Principal Components derived from each trajectory to assess convergence of sampling.
>
> Thank in advance for any guidance and advice
>
> Regards
>
> George
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
--
-------------------------
Daniel R. Roe, PhD
Department of Medicinal Chemistry
University of Utah
30 South 2000 East, Room 307
Salt Lake City, UT 84112-5820
http://home.chpc.utah.edu/~cheatham/
(801) 587-9652
(801) 585-6208 (Fax)
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Aug 05 2015 - 10:30:02 PDT
