What I usually do when I want an "average" structure is
to use MOIL-View to do a cluster anaalysis of the trajectory,
then have it save the most representative structure. This
is the one with the lowest average RMSD to all of the
others in the cluster. It makes it a physical structure that is
a good model for the ensemble that makes up that family.
MOIL-View does this automatically, other programs may do it too.
Carlos
David Konerding wrote:
> "Guanglei Cui" writes:
>
>> I think MoilView by Carlos Simmerling and ptraj by Tom Cheatham both can
>> generate average structures from dynamics. However, the averaged structure
>> can be distorted, especially when there's large conformational change such
>> as terminal base pair opening. I've seen the ring of one terminal base
>> shrinked significantly. Minimization doesn't help to correct this as far as
>> I know. Good luck!
>
>
> Right, imagine the case of a sugar pucker which repuckers between C2' endo and
> C3' endo. If the populations are 50% each, and you average them, you
> end up with a planar sugar pucker which is physically impossible. Minimization
> with pop the pucker down to the local minimum, but I doubt that's particularly relevant
> or useful.
>
> As for averaging, I usually use carnal or my own code. Once you've RMS aligned the
> structures to a common structure, you can just do a vector additon of the coordinate
> sets and then divide by scalar number of frames. Or, pre-divide and then sum.
--
===================================================================
Carlos L. Simmerling, Ph.D.
Assistant Professor Phone: (631) 632-1336
Department of Chemistry Fax: (631) 632-7960
SUNY at Stony Brook Web: http://comp.chem.sunysb.edu/carlos
Stony Brook, NY 11794-3400 E-mail: carlos.simmerling_at_sunysb.edu
===================================================================
Received on Thu Mar 29 2001 - 09:55:55 PST
