Re: [AMBER] bad atom type decomp

From: Jason Swails <>
Date: Tue, 5 Jul 2016 14:27:55 -0400

On Tue, Jul 5, 2016 at 10:04 AM, Bruno Falcone <>

> Hi, I was wondering if someone got a chance to look at my question.
> Speficically, if I go the recompilation route, what vdw radius should I
> use for iodine?

​The radii you refer to are used to define the solvent/solute boundary (it
is the radius of the iodine atom when defining the SASA). There are a lot
of precision issues with the non-polar solvation free energy model -- the
fit to the simple linear fit of nonpolar solvation energy to SASA model is
not particularly accurate (although not horrible), the effect of a single
iodine atom in the entire molecule will be pretty minimal, and a
"decomposed" surface area -- particularly for pairwise decomposition -- is
challenging to even conceptualize.

​So my advice is this: choose something that seems reasonable -- even if
there *is* some best definition of a "correct" choice here, if you pick a
radius that is reasonable then the effect of not choosing the "best" value
will be dwarfed by the effect of the other sources of error and
approximations here.


Note that what I'm saying here applies to single-point energy calculations
with a lot of pretty large approximations -- I'm certainly not suggesting
such a cavalier attitude
​to picking parameters for force fields in general.

Jason M. Swails
AMBER mailing list
Received on Tue Jul 05 2016 - 11:30:03 PDT
Custom Search