From: Jason Swails <>
Date: Fri, 20 Dec 2013 06:58:03 -0500

On Fri, Dec 20, 2013 at 2:23 AM, filip fratev <> wrote:

> Hi all,
> When I use only &pb in my MMPBSA calculations I obtain a reasonable value
> for EPB, but when I set idecomp=3 the EPB was about 10x smaller in the
> final_result_mmpbsa file (EPB 67.5297 ). I used the same input files and
> all other energies are indeed identical. Is that a bug? May I consider that
> the decomposition energies are correct?

PB is notoriously difficult to do any kind of pairwise decomposition.
 Since PB energies are not even remotely pairwise decomposable, anyway, the
pairwise residue energies are dubious at best.

Since neither PB or GB are pairwise decomposable, any pair-pair residue
energies are useful only for qualitative purposes. In this case, I would
suggest using GB instead of PB for decomposed energies because PB takes a
long time to compute decomposed energies whose origin is unclear to me
(unless a separate PB calculation is done for each residue pair leaving all
other residues uncharged?). GB pairwise energies come from a pairwise part
of the potential (but since they depend on the effective GB radius which is
not a pairwise quantity, GB cannot be formally pairwise decomposed).

If the PB decomposition is done as I suspect, it would explain why the
total EPB is very low but the decomposition energies are somewhat reliable.
 I would still suggest using GB for decomposition instead, and remembering
that the energies you get are only useful for relative comparisons.


Jason M. Swails
Rutgers University
Postdoctoral Researcher
AMBER mailing list
Received on Fri Dec 20 2013 - 04:00:02 PST
Custom Search