Re: [AMBER] Query about microsecond long simulation

From: Ross Walker <ross.rosswalker.co.uk>
Date: Wed, 29 Apr 2015 06:37:04 -0700

> On Apr 29, 2015, at 6:20 AM, anu chandra <anu80125.gmail.com> wrote:
>
> Dear amber users,
>
> I am working with membrane proteins and planing to do a microsecond long
> simulation. I have few queries for which I would like to get your expert
> suggestions.
>
> 1. I have seen from literature that many membrane protein simulations are
> carried out with Gromacs and NAMD engine. Since I am familiar with Amber, I
> wish to stick with Amber for the simulation. How safe to go with Amber for
> microsecond long membrane protein simulations?
>

Perfectly fine - we regularly run multiple microsecond long simulations in our lipid force field development work - and likely a lot more efficient (if you use GPUs) than you'd get trying to do this timescale with NAMD. You need to decide what force field you are going to use. I would suggest either the AMBER lipid 14 force field (and FF14SB for the protein), which is very easy to use with AMBER, or the latest Charmm lipid force field, although in this case you will need to convert the psf file with AMBER's chamber or parmed programs.

> 2. All my previous simulations were of 100-200ns. I ran the simulations
> with 5 ns window and saving the trajectories every 2 ps. Here, since the
> simulations are going to be microsecond long, How good to stick with the
> same protocol.?. I am very much aware of the large dataset going to be
> generated during this long run.
>

The frequency at which you write to the trajectory file is only an issue in terms of the amount of data you will produce. If you have plenty of disk space this is not an issue but you might want to consider making it less frequent if you don't need data every 2ps. Performance can be affected by writing too often but every 1000 steps+ is fine. 5ns windows is also fine, you might want to extend this to longer if that is easier for you to manage - I tend to try to shoot for 1 hour or so run time per simulation - but this is arbitrary and easy enough to script so just pick a time length per simulation that is convenient for you. Note you MUST set ig=-1 here so that the random number seed changes automatically on every step otherwise you will get unrealistic correlations in your results.

I'd also suggest using the GPU code otherwise it will take you a long time (and tons of resources) to get to microsecond time scales.

All the best
Ross

/\
\/
|\oss Walker

---------------------------------------------------------
| Associate Research Professor |
| San Diego Supercomputer Center |
| Adjunct Associate Professor |
| Dept. of Chemistry and Biochemistry |
| University of California San Diego |
| NVIDIA Fellow |
| http://www.rosswalker.co.uk | http://www.wmd-lab.org |
| Tel: +1 858 822 0854 | EMail:- ross.rosswalker.co.uk |
---------------------------------------------------------

Note: Electronic Mail is not secure, has no guarantee of delivery, may not be read every day, and should not be used for urgent or sensitive issues.


_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Apr 29 2015 - 07:00:03 PDT
Custom Search