[gmx-users] GROMACS reproducibility

[Justin A. Lemkul]

Lucio Ricardo Montero Valenzuela wrote:
> Hi. I am modelling a putative interaction between two proteins by doing MD of
> the docked complex with GROMACS 4.0. But each time I run the MD simulating 200
> ps, the evolves in a different manner, and the average coordinates are
> different in a loop believed to help to the interaction. When I plot the LJ and
> coulumb energies between the two proteins, they are two different. ¿Is it
> reasonable to select the simulation that gives the better energies and moves
> the interaction loop towards the interaction partner, and continue with these
> simulation for ~2 nanoseconds? ¿or I have to do further analysis?.

I suppose that depends on your intent.  Are you trying to reproduce 
experimentally-determined interactions that you know should occur?  Or do you 
have some other purpose?

In any case, you can't expect every simulation to do exactly the same thing. 
Only over long time periods will independent simulations converge in their 
observable properties.  Depending on the size of the loop and nature of the 
interactions between your proteins, 2 ns may not be enough.  Likely it is not, 
and given the speed of Gromacs 4, there is no reason that tens of ns can't be 
done in a reasonably short period of time.

-Justin

> Sincerely yours.
>      Lucio.
> 

-- 
========================================

Justin A. Lemkul
Ph.D. Candidate
ICTAS Doctoral Scholar
MILES-IGERT Trainee
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

========================================