Re: [AMBER] Problem with

From: Jason Swails <>
Date: Tue, 8 Jul 2014 08:44:03 -0700

On Tue, Jul 8, 2014 at 7:07 AM, Rajeswari A. <>

> Dear Amber developers,
> I am running version 12 (AMBER12 + Ambertools12). However, when I
> use PBSA solver in sander, the EPB term vary significantly with that of a
> run that used mmpbsa_py_energy, for the same trajectory.
> For example EPB is -2109.0597 with mmpbsa_py_energy and -41.7154 with
> sander.

​This should not happen. sander and mmpbsa_py_energy (and pbsa) use the
same PB code, and so should nominally give the same answer. There are some
small setup changes between the two programs that I've noticed, so the
energies don't match up exactly.

However, when I run the test case with sander and compare with the values I
get using mmpbsa_py_energy, the energies all agree to within 0.1 kcal/mol
(0.01 kcal/mol for binding, which is well inside the uncertainty of the
method itself).

What is your input file? It could be that is not writing the
input file for sander correctly. Although based on the sheer magnitude of
your numbers, I would guess that you are comparing the receptor
contribution calculated by mmpbsa_py_energy with the ligand contribution
computed by sander (-2109 is the right order of magnitude for a large
protein, while -40 is the right order of magnitude for a small ligand).

> Which one is better to use?

​Run the Amber test suite. Do all of the tests pass?


Jason M. Swails
Rutgers University
Postdoctoral Researcher
AMBER mailing list
Received on Tue Jul 08 2014 - 09:00:03 PDT
Custom Search