[gmx-users] Re: Representative structure

[Anton Feenstra]

Ing. Vojtech Spiwok wrote:

>>Hi all,
>>How can I get a "representative structure" of the end of a MD 
>>trajectory? Should I average the structures of the last time frames?
>>Thanks in advance,
>>Bests,
>>Ruben
>>
> 
> An averaged structure has incorrect stereochemistry. For example if
> methyl group rotates during the dynamics in averaged structure it would
> look like a upside down umbrella. On the other hand for example ligand
> docking to such structure might be succesful. Who knows.

I've used averaging over 100ps plus energy minimization succesfully for
the final optimization of a homology model. It performed significantly
better on (indeed!) ligand docking than the non-optimized structure.

If you need to be sure your structure is representative for your simulation,
you should probably to a cluster analysis, which can readily be done using
g_cluster. Depending on the characteristics of your system, it may take a
bit of fiddling with parameters to get useful results, but you will be
able to say how many % of your simulation is represented (in the major
cluster), and also what the other 'parts' of you simulation look like
(i.e., the other clusters).


-- 
Groetjes,

Anton
  _____________ _______________________________________________________
|             |                                                       |
|  _   _  ___,| K. Anton Feenstra                                     |
| / \ / \'| | | Dept. of Pharmacochem. - Vrije Universiteit Amsterdam |
|(   |   )| | | De Boelelaan 1083 - 1081 HV Amsterdam - Netherlands   |
| \_/ \_/ | | | Tel: +31 20 44 47608 - Fax: +31 20 44 47610           |
|             | Feenstra at few.vu.nl - www.few.vu.nl/~feenstra/         |
|             | "If You See Me Getting High, Knock Me Down" (RHCP)    |
|_____________|_______________________________________________________|