Re: [AMBER] positive Amber FF energies?

From: Thomas Cheatham <>
Date: Thu, 17 Dec 2020 04:29:17 +0000

Jason's reply is excellent as usual and also accurate, despite the fact that AMBER is now a side job for him and he is graciously answering on his own time. Absolute energies have little meaning. The +/- does not matter, only the relative.energies.

If you look at AMBER vs. CHARMM with solvated energetics, AMBER is more negative since CHARMM water models add a vdw on the hydrogen on their water (TIPS3P?) model making more vdw energy and positivity. Is it better/worse; look for the agreement to experiment. Yet, absolute energy values only matter when you can directly compare to an (exactly) equivalent system,

Details matter, as do assumptions. The assumption that negative is "favorable" (or required) is wrong - it is all relative...

We previously did direct potential energy comparisons; required equal # of waters, ions, etc.

From: Jason Swails <>
Sent: Wednesday, December 16, 2020 8:59:35 PM
To: AMBER Mailing List
Subject: Re: [AMBER] positive Amber FF energies?

On Wed, Dec 16, 2020 at 11:32 AM Vaibhav Dixit <>

> Dear All,
> I'm getting positive Amber FF energy values after I stripped the solvent,
> and ions.
> I did this for other proteins recently and got meaningful and -ve E values.
> Thus I'm a bit puzzled about what these energy values might mean.

You've hit the nail on the head. They mean absolutely nothing. The only
thing that has any meaning whatsoever are energy differences.

Constant shifts in the potential energy surface (for example, changing bond
potentials from [k*(x-x0)^2] to [k*(x-x0)^2 - 1e5] have no effect on forces
and, by extension, trajectories.

Also note that stripping the solvent and ions is a huge deal, and can
completely change whether a species is stable or not. As an example, the
dominant species in an ensemble of an isolated amino acid is its
zwitterionic form, where the amine group is positively charged and its
carboxylate negative. This flips in the absence of an aqueous solvent
(which works to stabilize those charges), and the dominant stable state in
the gas phase is its uncharged amine-carboxylic acid form.

>From the printout you posted below, it doesn't seem like you turned on any
kind of implicit solvent like GB, so I would not be at all surprised to
learn that many configurations are only stable because of the effects of
the solvent. But it's important to note that positive MM energies by
themselves are completely meaningless. In fact, the values you show seem
perfectly sensible. Bonds and angles are treated as ideal springs and so
cannot have an energy lower than 0, and dihedrals are frequently the same
way. The electrostatic interactions are slightly attractive, as are the
van der Waals (again, not surprising as the vdW potential is universally
attractive at longer distances and removing solvent and ions also removes
many potential close contacts where a positive vdW interaction was offset
by an attractive electrostatic one).

But for a solvated system, the bulk of the net electrostatic energy comes
from water-water interactions, which are highly favorable (and have a
negative value by virtue of the electrostatic potential energy function).
They don't contribute to bonded terms because their bonds are constrained
to their minimum-energy value (0), so they basically only contribute
nonbonded terms that tend to be negative. Ditto with the ions.

The above discussion highlights why positive values make sense in this
context. It's important to reiterate that the actual energy values in an
MM calculation are completely meaningless.

Visual inspection of the structure did not reveal anything strange and the
> structure looks fine.
> Thus can you please comment as to what might be wrong here and how can I
> possibly fix this problem?
> Sample Es for stripped protein
> 1 * 7.7864E+02 1.9139E+01* 1.1727E+02 FE 1491
> BOND = 284.3194 ANGLE = 828.9885 DIHED =
> 508.5032
> VDWAALS = -682.0879 EEL = -4975.7223 HBOND =
> 0.0000
> 1-4 VDW = 343.3292 1-4 EEL = 4397.5842 RESTRAINT =
> 0.0000
> CMAP = 73.7232
> minimization completed, ENE= 0.77863756E+03 RMS= 0.191394E+02
> TRAJENE: Trajectory file ended
> TRAJENE: Trajene complete.
> Sample E for solvated protein
> 1 * -6.1452E+04 1.3795E+01* 1.3013E+02 C 727
> BOND = 308.1534 ANGLE = 820.6275 DIHED =
> 504.6346
> VDWAALS = 8617.0697 EEL = -75800.8695 HBOND =
> 0.0000
> 1-4 VDW = 337.3554 1-4 EEL = 3694.9279 RESTRAINT =
> 0.0000
> CMAP = 66.4508


Jason M. Swails
AMBER mailing list
AMBER mailing list
Received on Wed Dec 16 2020 - 20:30:03 PST
Custom Search