[gmx-users] ED analysis, domain rotation, DynDom

Tsjerk Wassenaar tsjerkw at gmail.com
Sat Mar 8 11:49:02 CET 2008

Hi Jo,

First of all, you should note the difference between essential
dynamics analysis and sampling. Also note there's nothing essential
about it. But the first is done using g_covar/g_anaeig and the latter,
which concerns enhanced sampling along specific modes, is done using
make_edi/mdrun. So in your case, you want the first.

As to the question regarding the group for fitting, what you'll do if
you fit to one domain and calculate the covariance matrix for the
whole thing is, sort of, amplifying the inter-domain motion, while
suppressing the influence of the local motions. As the displacements
of the other (non-fitted) domain will be much larger, the covariances
involving these motions will also be the larger ones. Maybe this is
just what you want. It is explained in a (little) bit more detail in
our Proteins paper from this month.

Finally, whether to use deviations from a reference are from an
average. I'd say the results are better defined if you use deviations
from the average, as you'll be talking about central moment principal
components. Non-central principal component analysis requires a lot
more thought and explanation.

Try to read some statistics literature on principal component analysis
to get more feel for the method (and at any point realize that wer'e
so lucky to be able to give physical meaning to the components we get

Hope it helps,


On Fri, Mar 7, 2008 at 11:50 PM, jo hanna <jfhanna at gmail.com> wrote:
> Hi
> I have a done number of MD simulations of a two domain protein with
> different ligands and also in the apo state. Comparing these simulations I
> see varying degrees of domain rotation of the one domain relative to the
> other which I want to quantitatively measure and I was thinking of using
> DynDom to do this; I have already assessed a number of structures taken from
> different points throughout the simulations but want a to assess the motion
> throughout the simulations.
>  >From the user lists I read that I should do ED analysis first. Is this
> using g_covar and then g_anaeig or using make_edi and then doing ED in mdrun
> and then assessing these results?
> Also as the motion appears to be the C-terminal domain relative to the
> N-terminal domain when using g_covar should I fit over the N-terminal domain
> backbone but do the analysis over the whole backbone, or just fit over the
> whole backbone? Also in g_covar is it better for this type of problem to use
> the -ref option and assess the deviation from the conformation in the
> structure file instead of from the average?
> I would appreciate some guidance as although I've read some papers on this
> subject I am a bit confused as to what is the best way to progress.
> Many thanks,
> Jo.
> _______________________________________________
>  gmx-users mailing list    gmx-users at gromacs.org
>  http://www.gromacs.org/mailman/listinfo/gmx-users
>  Please search the archive at http://www.gromacs.org/search before posting!
>  Please don't post (un)subscribe requests to the list. Use the
>  www interface or send it to gmx-users-request at gromacs.org.
>  Can't post? Read http://www.gromacs.org/mailing_lists/users.php

Tsjerk A. Wassenaar, Ph.D.
Junior UD (post-doc)
Biomolecular NMR, Bijvoet Center
Utrecht University
Padualaan 8
3584 CH Utrecht
The Netherlands
P: +31-30-2539931
F: +31-30-2537623

More information about the gromacs.org_gmx-users mailing list