Thanks this is great, I'll try your suggestions.
So you basically suggest to run a series of chunks of MD within the same script?
Could you please send me a sample script like this?
And use ntp=0 for constant volume only?
Thanks!!
Fabian
_______________________________
Fabian Glaser, PhD
Bioinformatics Knowledge Unit,
The Lorry I. Lokey Interdisciplinary
Center for Life Sciences and Engineering
Technion - Israel Institute of Technology
Haifa 32000, ISRAEL
fglaser.technion.ac.il
Tel: +972 4 8293701
Fax: +972 4 8225153
On Nov 29, 2012, at 5:04 PM, Jason Swails wrote:
>
>
> On Thu, Nov 29, 2012 at 8:44 AM, Fabian Glaser <fglaser.technion.ac.il> wrote:
> Works!!
>
> I am running production dynamics as in the tutorials (first time), so allow me to ask you if this settings are good for a medium size protein (130 aa) with 6000 waters. For 30 ns ....
>
> production dynamics
> &cntrl
> imin=0, irest=1, ntx=5,
> nstlim=15000000, dt=0.002,
> ntc=2, ntf=2,
> cut=10.0, ntb=2, ntp=1, taup=2.0,
> ntpr=1000, ntwx=1000, ntwr=50000,
> ntt=3, gamma_ln=2.0,
> temp0=300.0,
> /
>
> A few comments here:
>
> 1) Constant pressure MD is noticeably slower than constant volume MD (since NPT requires calculating the virial). While you should always use NPT at some point in your simulation, I typically restrict this to my equilibration stage and wait until the density (or box size) stabilizes. Once this occurs, NPT and NVT are more or less indistinguishable in aqueous solvent (since water is approximately incompressible).
>
> 2) I discourage you from trying to run something like 30 ns of simulation at once (unless you're simulating <100 atoms or something). I typically run simulations in chunks of 1 or 2 ns. The benefit of this approach is that you can submit shorter jobs which are more likely to start in most clusters. You can also submit the jobs in such a way that you submit 30 1-ns jobs at once, and each one is held from running until the job in front of it finishes. This kind of approach is far less susceptible to being killed by a node failure and will likely finish much faster.
>
> For an introduction on using PBS effectively, see http://jswails.wikidot.com/using-pbs
>
> Good luck,
> Jason
>
> --
> 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 Sun Dec 02 2012 - 05:30:02 PST
