[gmx-users] implicit solvent protein simulations with charmm

[Mark Abraham]

On 14/08/2012 1:07 AM, Sai Pooja wrote:
> Hi,
>
> I have read a couple of posts on related topics but I am still not
> sure how to proceed and would be grateful to get some help.
>
> I want to run implicit solvent simulations of a protein+water
> (implicit) system in the canonical ensemble using charmm27+cmap
> forcefield. I am trying to use the parameters used in the 2010 JCTC
> paper on the implementation of charmm in gromacs. In this paper, it
> seems that Stochastic Dynamics with zero cut-off is used for the
> implicit solvent system. Moreover, the inverse friction constant is
> specified as 91 ps-1. It is not clear to me how does one specify this
> inverse friction constant with the "sd" integrator in gromacs.

Manual 7.3.3 says how.

>   Does it
> correspond to "mass/tau_t" or the bd_fric (which I understand is used
> with the "bd" integrator)? If this is infact equal to mass/tau_t, what
> tau_t value does it correspond to (18/tau_t = 91)? The paper also
> claims to be able to use a time step of 5fs with OBC+virtual sites.
> Again, I would like to find out the right set of parameters that would
> allow this time-step.

If you contact the authors, they may be willing to share an .mdp file 
with you. Even if they do, you doing some careful reading of manual 7.3 
is in order. Attempting to reproduce some old results is likely to be a 
good idea.

Mark