Re: [AMBER] Large different between experimental and MM-GBSA absolute energy, any potential source of error?

From: Jason Swails <>
Date: Mon, 21 Feb 2011 12:58:23 -0500


MMPBSA makes a lot of very large assumptions. First, configurational
entropy is often neglected, and even with a normal mode approximation you
may still be missing large parts of the entropic contributions. Secondly,
the "ligand" and "receptor" often have to adopt slightly strained geometries
in order to properly bind. However, MMPBSA often incorporates the
assumption that bound and free trajectories are the same, which often
significantly underestimates the stability of either free species (forcing
them to adopt higher-energy structures). Then, when binding free energies
are calculated in MMPBSA by subtracting the ligand and receptor values from
the complex value, the ligand and receptor values are most likely going to
be artificially high (best case scenario they are the *correct*
trajectories, but this is unlikely), causing your binding free energy to be
significantly more negative than it would otherwise be. Entropy is also
often underestimated and frequently destabilizes a binding interaction.

For these reasons (and a few others), MM/PBSA is not a reliable technique
for calculating absolute binding free energies. However, for relative
binding free energies these assumptions are made in every system, and if the
systems are similar you can make the argument that these errors largely

Hope this helps,

On Mon, Feb 21, 2011 at 5:38 AM, Catein Catherine <>wrote:

> Dear Sir/Madam,
> I have done a caloritm measurement on a binding between two proteins, and a
> protein with DNA.
> Experimentally, they are in the found to be -10 and -16 kcal/mol,
> respectively.
> I also do the MM-GBSA binding energies for the pairs of data, however, they
> were found to be -61 and -120 kcal/mol.
> Is it well know that MM-GBSA tends to overestimate the binding energies?
> Can I rationizal the enhanced binding energies or give reasons why they are
> different so large?
> I thought it could be due to salt was added in the experiment, ~ 0.2M NaCl.
> So, I add salt concentrate to ~ 0.2M. However, the theoretically energy
> only down to -50 and -100 kcal/mol, still far away from the experimental
> results. Any suggestions on the source of error?
> Best regards,
> Catherine
> _______________________________________________
> AMBER mailing list

Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
AMBER mailing list
Received on Mon Feb 21 2011 - 10:00:04 PST
Custom Search