Re: [AMBER] positive free energy

From: Ray Luo, Ph.D. <ray.luo.uci.edu>
Date: Wed, 20 Aug 2014 12:21:13 -0700

I think one reason is that the original inp=2 method was calibrated in
reproducing relative affinities due to small changes, i.e. side chain
mutations and/or different ligands.

If you use it to model absolute binding affinities between proteins
and nucleic acids, it breaks down. So neither inp=1 nor inp=2 gives
"correct" absolute binding affinities when the "ligands" are also
macromolecules. Nevertheless, I still think the relative affinities
are better than inp=1 if the relative changes are small. So far, I
haven't seen any data on this mailing list that show that the relative
affinities by inp=2 are worse than those by inp=1.

To address those large ligands, a different calibration has to be
developed, i.e. solute scaling now matters and should be taken into
account in the model. We are looking into improving the model for the
next release.

All the best,
Ray

--
Ray Luo, Ph.D.
Professor,
Biochemistry, Molecular Biophysics, and
Biomedical Engineering
University of California, Irvine, CA 92697-3900
On Wed, Aug 20, 2014 at 11:34 AM, Jason Swails <jason.swails.gmail.com> wrote:
> On Wed, 2014-08-20 at 11:11 -0700, zahra khatti wrote:
>> Dear Amber users
>>
>> I obtained positive free energy from mmpbsa.py.
>
> MM/PBSA absolute binding free energies are frequently unreliable.  It is
> useful mainly for rank-ordering binding affinities for related systems.
>
>> In my case the receptor and ligand didn't joint together. The distance between them
>> is about 1-2 angstrom. Is this  related to positive free energy?
>
> Let's have a look:
>
> [snip]
>
>> Differences (Complex - Receptor - Ligand):
>> Energy Component            Average              Std. Dev.   Std. Err. of Mean
>> -------------------------------------------------------------------------------
>> VDWAALS                    -50.0785                1.0295              0.0084
>> EEL                         -1.1552                0.3301              0.0027
>> EPB                          7.3643                0.8011              0.0065
>> ENPOLAR                    -22.2317                0.3450              0.0028
>> EDISPER                    354.9084                4.6963              0.0383
>
> Yikes.  The EDISPER component of your binding free energy is +350
> kcal/mole.  Combining that with the repulsive ENPOLAR term, that means
> your _total_ non-polar solvation free energy contribution to binding is
> about +330 kcal/mole, which seems outrageous to me.
>
> It seems to me that this is a fairly frequent occurrence with the
> "inp=2" non-polar solvation free energy model, so I can't say I
> recommend using it.  I suggest adding inp=1 to the &pb section of your
> input file to use the more prototypical, simple SASA-based nonpolar
> solvation approximation.
>
> .Developers: Perhaps we should consider either turning the default back
> to inp=1 or removing the default choice altogether and requiring users
> to specify it?  inp=2 seems to be causing a lot more problems than it
> solves and gives ludicrous answers more often than inp=1.
>
> All the best,
> Jason
>
> --
> Jason M. Swails
> BioMaPS,
> Rutgers University
> Postdoctoral Researcher
>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber

Received on Wed Aug 20 2014 - 12:30:02 PDT