Re: [AMBER] MMGBSA & igb 8

From: Carlos Simmerling <>
Date: Mon, 20 Oct 2014 10:43:06 -0400

I agree with Jason and would just add that I usually encourage people to
follow protocols that they see in the literature for applications similar
to their own. We have not tested igb=8 (GBneck2) with MM-GBSA, and it did
not have any small molecules in the training set. It may work well, it may
not - this would need testing. We have done lots of training and testing on
protein dynamics, which was the goal of the GB method. I would not really
know what to expect for ligand binding. If you're doing MM-GBSA on
protein-protein interactions, or protein-peptide, then it is withing the
design of the model, otherwise if it's small molecule binding then it's
outside the design and testing of the model. Others may have published
tests of GBneck2 for ligand binding, but I don't recall seeing any. If so,
look and see what options they used.

So overall, I suggest using a model and parameters that have been tested
for the application you want, or be prepared to try some different
parameters and compare against experimental data on your own.
best regards,

On Mon, Oct 20, 2014 at 10:34 AM, Jason Swails <>

> On Mon, Oct 20, 2014 at 6:01 AM, <> wrote:
> > Dear Users,
> > Is there any recommended value for SURFTEN and SURFOFF terms if one
> > would like to use igb 8 implicit solvent model for MMGBSA calculation?
> >
> ​It's no different than the values you'd use for other GB models to my
> knowledge. The GB model only specifies how the polar solvation free energy
> is calculated. The surften and surfoff parameters control the model for
> the nonpolar solvation free energy.​
> HTH,
> Jason
> --
> Jason M. Swails
> BioMaPS,
> Rutgers University
> Postdoctoral Researcher
> _______________________________________________
> AMBER mailing list
AMBER mailing list
Received on Mon Oct 20 2014 - 08:00:03 PDT
Custom Search