Re: [AMBER] Significantly different results between MMPBSA and MMGBSA

From: Jason Swails <>
Date: Sat, 10 Jan 2015 23:04:21 -0500

On Sat, Jan 10, 2015 at 10:55 PM, asdda asa <> wrote:

> Dear Amber users,
> I have problem with the result of MMPBSA and MMGBSA calculation.
> Previously, I performed three md simulations, each consisted of a ligand
> bound to the same protein. The three systems differ in ligand structures
> only.
> I noticed that there were significantly different results of binding
> energies calculated by MMPBSA and those by MMGBSA.
> The final binding energies calculated by GB were about -50 kcal/mol for
> each system, while those are from PB were about +10 kcal/mol for each
> system.
> Please anyhelp how could these difference occured?

​Look at the breakdown of the binding free energy contributions. I suspect
that the nonpolar free energy contribution (ENPOLAR+EDISPER for PB and
ESURF for GB) are very different from each other. They use different
models by default. I suspect that you can account for almost the entire 60
kcal/mol difference between PB and GB binding energies simply through this

It's unclear to me which is "better", but in this case I would probably go
with the simple model used by GB. You can use this model in the PB code by
setting inp=1 in the &pb section of the input file.

​Also make sure you are using the most up-to-date version of AmberTools.


Jason M. Swails
Rutgers University
Postdoctoral Researcher
AMBER mailing list
Received on Sat Jan 10 2015 - 20:30:03 PST
Custom Search