Hello,
On Mon, May 9, 2011 at 9:46 PM, Bozell, Joseph John <jbozell.utk.edu> wrote:
> I have successfully been carrying out MD simulations of an array of 10 self
> assembling molecules within an explicit water box at relatively short
> simulation times. They appear stable, so I have set up a significantly
> longer (for us) run of 5 ns on our university's mainframe. Looking ahead, I
> know this will generate a large .mdcrd file even with ntwx set higher (5000
> steps) and the netCDF flag set.
>
The definition of "large" certainly seems to be relevant to average disk
size and processor speed. Indeed, 5 ns doesn't seem so large in current
standards (given a *normal* time step and system size).
>
> In the past, I have been able to visualize the simulation using VMD on my
> Mac. However, if the files get much larger than 750-800MB, my computer seems
> to choke and is unable to complete the visualization. Are there any
> recommendations for visualizing larger runs? How are production runs going
> into the tens of ns effectively managed (filesize-wise) and visualized?
>
Well you definitely want to use strictly NetCDF files; that's a no-brainer.
On top of that, there's really only one way I can see to reduce the file
size short of increasing ntwx, and that relies on you having a system with
explicit water (if you don't, then this likely doesn't apply).
If you do have an explicit solvent calculation and you don't care about the
coordinates of the water, then you can just create a topology file *without*
any water and do all of your visualization and analysis with that. In fact,
you can use the input variable ntwprt to control how many atomic coordinates
are printed out (every x, y, and z coordinate of the first ntwprt atoms are
printed to the trajectory file).
Remember, however, that you need to create a compatible topology file to do
anything with these output trajectories.
You can also post-process the trajectories after-the-fact and strip out the
solvent using ptraj (the compatible topology comment still applies for
visualizing this trajectory, obviously).
HTH,
Jason
> Thanks much,
>
> Joe Bozell
> University of Tennessee
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
--
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
352-392-4032
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Mon May 09 2011 - 15:00:04 PDT
