Re: [AMBER] Regarding high temp simulation of DNA and amber MD module

From: David A Case <>
Date: Sat, 23 May 2015 12:38:54 -0400

On Sat, May 23, 2015, MOHD HOMAIDUR RAHMAN wrote:

> So my concern is about amber simulation module. Earlier in my lab, we are
> following a protocol that is based on sander.mpi module of AMBER. Now I am
> planning to run fresh all system on pmemd.mpi or pmemd.cuda.mpi. So how my
> result will affect due to shifting from sander.mpi to pmemd.mpi or
> pmemd.cuda.mpi. (In term of non-bonded interaction). Which module is
> optimal in case accuracy and speed. I am using full ewald method for
> calculating electrostatic interaction with cutoff 15.0.

To add to Carlos' answer: sander and pmemd and pmemd.cuda all all implement
the same force fields, and give equivalent results. I'd strongly recommend
using PME with the default cutoff as you get started. pmemd (for CPU)
or pmemd.cuda (for GPU) will give the best performance.

If you are convinced you need "regular" Ewald (instead of PME), you have to
stick with sander. But you almost certainly want PME instead.

The Amber web site has a tutorial on ionic liquids; adding DNA to what is done
there is not a gigantic change. Be sure to consider whether relevant
experiments have any water or alkali ions in addition to the ionic liquid
components, since that could have a big effect on the results.


AMBER mailing list
Received on Sat May 23 2015 - 10:00:03 PDT
Custom Search