Re: different seeds and protein behavior

From: Dr. Eddie (eackad_at_gmail.com)
Date: Fri Jul 20 2012 - 07:56:44 CDT

Thanks you all for your input. If I may ask a more specific followup: Isn't
it entirely possible (if not likely) that a protein can have multiple
convergent behaviors?

Say I am interested in the distance between two residue's carbon-alpha's,
but assume the protein will not denature and is started close to its 0K
structure. I perform 10 runs and I get two distinct behaviors (one distance
distribution around 6 Angstroms the other around 8). Which is more likely:
1) The protein has two distinct "heated" dynamical behaviors
or
2) I have not run for a sufficient amount of time and I should expect with
more time each type will "switch" over and in the end I will get the
average.
Thanks,
Eddie

On Fri, Jul 20, 2012 at 3:14 AM, Nicholas M Glykos <glykos_at_mbg.duth.gr>wrote:

>
>
> What you refer to is the crux of sufficient sampling. If different runs
> (or different parts of the same run) produce statistically different
> results, then your measured quantites have not converged (and you should
> continue the simulation). Although demonstrating sufficient sampling is
> the name of the game in the field, in some cases (eg folding simulations)
> there is not a consensus of how to demonstrate it in the first place
> (eigenvector overlap from dPCA or cPCA is our current favorite, but
> opinions across the community do vary).
>
> My twocents,
> Nicholas
>
>
> --
>
>
> Nicholas M. Glykos, Department of Molecular Biology
> and Genetics, Democritus University of Thrace, University Campus,
> Dragana, 68100 Alexandroupolis, Greece, Tel/Fax (office) +302551030620,
> Ext.77620, Tel (lab) +302551030615, http://utopia.duth.gr/~glykos/
>
>

-- 
Eddie

This archive was generated by hypermail 2.1.6 : Tue Dec 31 2013 - 23:22:17 CST