Re: [AMBER] idecomp and so on....

From: Jason Swails <>
Date: Tue, 9 Oct 2012 23:11:49 +0200

On Tue, Oct 9, 2012 at 10:59 PM, "André C. Stiel" <> wrote:

> Wow, thanks a lot, it works!
> Nobox, Reading that part of the ptraj manual would have actually
> helped, ehm!
> Could you comment shortly on my parameters: igb = 5, gbsa = 2 does
> it make sense with the afore mentioned
> run parameters (ntb=2/ntp=1 ...) or are there possible pitfalls
> involved?

There are always pitfalls ;). Keep in mind that the decomp energies are
potentially misleading. Regardless of your energy function, energies are
not pairwise-decomposable (for GB, the polar solvation free energy depends
on the GB radii for each atom, which are calculated as a sum over every
other atom; and the non-polar free energy based on the SASA is also a
non-local, non-pairwise-decomposable number).

For PB, the dielectric boundary is also based on every atom. Finally, even
for PME, the electrostatic energy is not pairwise decomposable since the
reciprocal space sum is not pairwise decomposable. It can be qualitatively
useful in a comparative way. In this case, I expect you'll see little
difference between different GB models or gbsa=0/2.


Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
AMBER mailing list
Received on Tue Oct 09 2012 - 14:30:03 PDT
Custom Search