[gmx-users] Different initial conditions for the membrane protein simulation

[Mark Abraham]

On 11/09/2012 2:36 AM, James Starlight wrote:
> Dear Gromacs Users!
>
>
> I'm simulating intrinsic dynamics of some membrane receptors which can
> have functional relevance. Therefore I'm looking for the most suitable
> initial conditions of my simulation which could provide maximal degree
> of such dynamics (arisen from the cooperative thermal fluctuations for
> instance). As I know the proper chosen thermostat can significantly
> influence on the overall dynamics of the protein. In particular I've
> found that Noose-Hover thermostat can promote it significantly. But
> there are some theoretical studies which suggest that Nose-Hover
> thermostat can produce alot of artifacts due to the incorrect energy
> distribution on the micro-states. So the dynamics produced onto such
> trajectories can be unphysical in general. That authors also suggest
> use collision thermostat ( in GROMACS usage of SD integrator
> correspond to that conditions without any t_coupl ).

Or the V-rescale thermostat, which claims correct thermodynamics and 
kinetics, and is much easier to use than N-H.

> So the main question- IS there any way to observe impact of each
> thermostats and other initial conditions on the overall dynamic
> behavior of the protein. For example does it possible to make some
> energy landscape of the trajectories calculated from different
> conditions to compare thermal energy distribution on the micro-states?
> Finally is there any way to compare such energy landscapes with the
> conformational maps ( e.g projections onto some lowest Principal
> components) to direct monitor the energy contribution of each
> conformation ?

Not sure. The difficulty with comparing multiple setups is that in 
principle you need a converged simulation for each and some way to 
assess which is the best model of reality, and usually at least one of 
those is infeasible.

Mark