[gmx-users] Density of water, Dispcorr and rmsd

[Mark Abraham]

On 9/03/2011 10:43 PM, Kavyashree M wrote:
> Dear users,
>
>     What is the difference between densities predicted by tip4p and 
> spc models?
> I am using tip4p (with OPLSAA forcefield) in the simulation, during 
> position restrained
> dynamics, Average water density is around 1015kg-m-3 at 300K which is 
> quite far from
> expt value..

You're modelling a flexible three-atom two-lone-pair fluid that does 
proton transfer and hydrogen bonding with a non-dissociative rigid 
four-point three-parameter model. 1.5% error starts to sound OK 
actually. If you were actually using position restraints on those water 
molecules, then all you are observing is the density of the starting 
configuration. Anyway, deviations of up to about 20-30 are expected, and 
the size and sign depend on your pressure coupling scheme, the water 
model, and what the water model was parametrized to reproduce.

>
>     The Dispersion correction (DIspCorr) is applicable only when we 
> use vdw_type = cut off?
> Any known disadvantage of using vdw_type = switch?

Please do your homework reading the (for example) the manual and primary 
literature, as suitable.

>     What could be the possible reasons for getting quite noisy and not 
> equilibrating RMSD
> of backbone  during MD simulations?

Non-ergodicity of finite simulations, a broken model, an intrinsically 
unstable system, wrong observations, periodicity artefacts, lack of 
experience in judging how significant noise is... anything really :-)

Mark