From: Carlos Simmerling <>
Date: Tue, 24 Oct 2017 09:51:48 -0400

I think it's a good idea to do what you mentioned, look at articles that
have tested the methods for performance on problems similar to yours.
However, as a GB developer, my opinion is that GB is really very
approximate and I only use it where speed is most important (such as in
MD). In postprocessing, speed is probably much less important than
accuracy, and in those cases I opt for PB over GB. Also keep in mind that
there are many different GB models, and their performance might differ for
your problem. For example, when we developed the GBneck2 model in Amber
(igb=8), we optimized for proteins and nucleic acids, but NOT for small
molecule ligands. I would not expect it to perform as well as PB since no
data of that type were included in the training or the testing.

On Tue, Oct 24, 2017 at 9:47 AM, sangita kachhap <>

> Dear Amber developer
> I have to compare the binding free energies of different small molecule
> ligands with a protein. According to some of the research article GBSA
> method is good in ranking the free energies of ligands than PBSA. So which
> method is actually more reliable for comparing the binding free energies of
> ligands.
> If difference of binding free energies of two ligands is 5-6 kca/mole. Are
> these methods able discriminate the such energy difference?
> Thanks
> _______________________________________________
> AMBER mailing list
AMBER mailing list
Received on Tue Oct 24 2017 - 07:00:04 PDT
Custom Search