Re: [AMBER] Comparison of the results

From: Jason Swails <>
Date: Thu, 8 Mar 2012 17:00:20 -0500

On Wed, Mar 7, 2012 at 5:23 PM, manikanthan bhavaraju <> wrote:

> Hi All,
> I have performed thermodynamic integration using amber8 anf ff99, I
> have calculataed deltaA for protein-ligand system. Later with
> Ambertools 1.5 ( MM-PBSA, MM_GBSA, and NAB were done. When
> the TI results are compared with GBSA/PBSA +NAB, I am getting
> conflciting results.

To clarify, I think by "NAB" you mean normal mode, or Nmode, yes? NAB is
actually a programming language that can be used to do GBSA, PBSA, and
normal mode calculations (in fact, uses a nab program to do its
standard GBSA and PBSA calculations, as well as its normal mode

> TI results deltaA = -9.2 kcal/mol

Are these results converged? If so, this is a better estimate than you
could expect to get with MM/PB(GB)SA analyses.

MM-PBSA = -19.27 +/- 3.09 kcal/mol
> MM-GBSA = -11.85 +/-2.34 kcal/mol
> NAB (entropy) = -19.59 +/- 4.16 kcal/mol (10 snapshots)
> The PBSA and NAB values are nearly equal. But when I take the
> difference between GBSA and NAB values, the deltaG= +7.74 kcal/mol.
> There are no experimental values avaliable for my system for
> comparison.

How were the snapshots chosen? Are you using only a single trajectory?
 Are you sure the snapshots are uncorrelated?

The literature regarding MM/PBSA may have helpful insights (regarding
MM/PBSA shortcomings, particularly about the entropy contributions) as well.

I have done some literature search on free energy of binding. In some
> of the papers, the GBSA/PBSA were lower than NAB/nmode values. Are my
> NAB values meaningful?

This is a loaded/controversial question.

Any suggestions?

MM/PBSA is most useful for relative binding free energy calculations and is
far less successful for absolute binding free energy calculations. If you
have the TI results, I would trust those a priori. However, TI results are
often very expensive to really converge. That's the main reason why
MM/PBSA and MM/GBSA analyses are still widely used.


Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
AMBER mailing list
Received on Thu Mar 08 2012 - 14:30:04 PST
Custom Search