Re: [AMBER] MMPBSA warning : No skew or curtosis when zero variance in moment

From: Jason Swails <>
Date: Mon, 17 Nov 2014 08:57:32 -0500

On Mon, 2014-11-17 at 11:53 +0330, Mina Madah wrote:
> Dear all
> I am performing mmpbsa calculation for DNA-ligand (amber 10). I face whit
> two problem through running and analyzing:

I suggest upgrading to AmberTools 14, which provides better error
messages and is overall better supported.

> 1. when running, this massage is appeared:
> No skew or curtosis when zero variance in moment
> No skew or curtosis when zero variance in moment
> No skew or curtosis when zero variance in moment

The kurtosis and skew are the 3rd and 4th moments of a distribution (the
mean and variance being the first two). If the variance is zero (which
happens when, for instance, you only have one data point or all data
points are the same), then all moments after the variance are zero as
well, and every value is exactly equal to the mean (i.e., the
distribution function is a delta function).

> after termination, and get the snapshot_statistics.out, I check the
> snapshot_com.all.out,
> no massages or strange word is observed. By watching binding_energy.log file,
> this warning is written in it:
> Reading snapshot_rec.all.out
> Checking consistency of CALC
> Checking consistency of DATA
> Entry 1 NM SROT 0 exists -> overwriting
> Entry 1 NM SROT 1 exists -> overwriting
> Should I concern or ignore this warning? and how can I fix it?

This is one example of where the error message is improved. You are
pointed to this website for this error message:

> 2. Energy decomposition in snapshot_statistics.out shows unreliable data such
> as:
> VDW energy is positive, and INT energy is not zero?
> system consists of DNA, ligand and Na+ ions. Simulation is done in explicit
> solvent for 5 ns.

I fail to see why this data is "wrong" or "unreliable". It's perfectly
possible for the net vdW interactions between two residues to be
positive if there is a steric clash (that would be offset by other more
attractive terms). Also, certain internal energies *will* be non-zero
depending on your simulation settings, but you never told us what kind
of decomposition calculation you were running or which values turned out
positive, so I can only speculate.

I would suggest upgrading to AmberTools 14 and using, which I
(and others that respond on this list) am better able to help with.
And, as always, provide more information about *exactly* what you have
done and the *exact* problems you are having so we don't have to guess
those details.


Jason M. Swails
Rutgers University
Postdoctoral Researcher
AMBER mailing list
Received on Mon Nov 17 2014 - 06:00:03 PST
Custom Search