[gmx-users] box size with pressure coupling

[Erik Lindahl]

"Lynne E. Bilston" wrote:

> Hi all,
>
> Maybe I'm missing something obvious here, if so please enlighten me.
>
> I'm doing some pressure coupled simulations of a lipid/protein combination,
> starting with the usual em, then pr runs.  I follow that with a bit of
> equilibration with no restraints, using pressure coupling to zero pressure,
> to let the box size come to some equilibrium state. That's all fine. Then I
> use the output from that run to start another pressure coupled run, this
> time a semiisotropic pressure coupled run.
>
> The weird thing is that despite using the coordinates (-c pdb file) and the
> trajectory ( -t trr file) from after the  equilibration, the box size at
> the start of the second pressure coupled run resets itself to the size
> before the original pressure coupled run. Is grompp taking the coordinates
> (and thus box size) somehow from the first frame of the trajectory instead
> of the last as specified in the manual?  To summarise, I am doing the
> following:
>
> 1. em (OK)
> 2. pr (OK)
> 3. equilibration - md with isotropic pressure coupling, p=0 (runs OK)
> 4. md with semi-isotropic pressure coupling, resets box size at the start
> of the simulation to size after step 2.
>
> I checked that the box size is correct in the pdb file I'm using as input
> for grompp in step 4, so I guess it must be getting the initial coordinates
> and size from the trajectory file, not the finishing coordinates?? This
> only seems to happen when I use the trajectory option, not the coordinates
> file alone. I just wanted to keep the velocities etc.

Hi Lynne,

You can choose which frame in the trajectory to use with the -time option to
grompp, but it *should* read the last frame unless you specify anything. Maybe
we have a bug here!

Could you use the command 'gmxdump' to check the initial box dimensions in the
.tpr file you create with grompp?

e.g.

gmxdump -s topol.tpr | more

and then search for 'box' by pressing '/' , then 'box' + return.


By the way - I usually find it simpler to use the .gro format instead of pdb;
in this case the velocities are stored in the same file (with 3 decimals of
accuracy) as the coordinates and you don't have to care about using the
trajectory at all...

Cheers,

Erik