[gmx-users] Normal mode analysis after equilibration

[Mario Fernández Pendás]

Dear David,

Thank you very much for your answer.

2016-01-14 20:20 GMT+01:00 David van der Spoel <spoel at xray.bmc.uu.se>:

> On 14/01/16 18:23, Mario Fernández Pendás wrote:
>
>> Dear All,
>>
>> I am running simulations of a toxin placed in the middle of a bilayer. It
>> is known from the experiments that the equilibrium point of the toxin is
>> in
>> one of the surfaces of the membrane.
>>
> What is an "equilibrium point"?
>

I mean the position where the toxin tends to move.

I run MD with v-rescale a number of time-steps big enough to get the system
>> equilibrated with the toxin in the proper place.
>>
>> At this stage, if I try to construct the Hessian matrix aiming to get its
>> eigenvalues without any energy minimization, it is likely that such
>> eigenvalues will be negative, isn't it? Even considering smaller
>> time-steps
>> or tolerance?
>>
> Yes you are bound to get negative eigenvalues.
>
> Why would you want to do NM of part of a complex system anyway? You will
> suddenly throw away all the correlations with the membrane and solvent.
>
> In my opinion normal mode analysis virtually only makes sense for isolated
> systems in the gas-phase
>
> My intention was to try to observe how the frequencies of the different
modes change after the simulation and how different are them with respect
to the ones associated to the harmonic bonds of the system in equilibrium.
That is why I started to consider studying the eigenvalues of the Hessian.

Thank you very much.
Cheers,
Mario