Re: AMBER: Thermodynamic Integration for a CHARGED molecule

From: David A. Case <case.scripps.edu>
Date: Fri, 17 Aug 2007 08:46:38 -0700

On Fri, Aug 17, 2007, David Mobley wrote:
>
> There are two basic approaches to this problem, which is a tricky one:
> (a) Use a treatment of long range electrostatics that will handle this
> for you.

The basic paper on the Amber approach to this is here:

%A T. Darden
%A D. Pearlman
%A L.G. Pedersen
%T Ionic charging free energies: Spherical versus periodic boundary
conditions
%J J. Chem. Phys.
%V 109
%P 10921-10935
%D 1998

Note that many of the "corrections" for net charge that are mentioned in the
literature refer to PME or Ewald implementations that differ from Amber's:
these "corrections" are already included in the way we do things.

As David notes, even carefully defining the absolute charging energies is
tricky, although I think he overstates the difficulties. For relative
charging free energies (which is what I think you were talking about),
uncertainties in how to handle the net charge effect should cancel.

More discussion of this general area is given here:

%A T. Simonson
%T Electrostatics and dynamics of proteins
%J Rep. Prog. Phys.
%V 66
%P 737-787
%D 2003

%A T. Simonson
%A J. Carlsson
%A D.A. Case
%T Proton binding to proteins: pKa calculations with explicit and implicit
%solvent models
%J J. Am. Chem. Soc.
%V 126
%P 4167-4180
%D 2004

%A M.A. Kastenholz
%A P.H. Hunenberger
%T Influence of artificial periodicity and ionic strength in molecular
dynamics simulations of charged biomolecules employing lattice-sum methods
%J J. Phys. Chem. B
%V 108
%P 774-788
%D 2004

In my view (unproven!), the biggest methodological problems are likely to come
about by using too small a periodic box, i.e. by "artificial periodicity",
rather than by having a non-zero net charge. It is possible that errors from
too small a box size could be reduced by charging an explicit counterion to
maintain neutrality, but I haven't seen any studies that speak to this point.

Also, in my experience, the errors that may arise from lattice-sum models are
completely overwhemed in proteins by the failure to correctly sample all of
the relevant conformations, and by errors in the force fields for these
complex systems.

...regards....dac

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Sun Aug 19 2007 - 06:07:41 PDT
Custom Search