From: "Xavier Periole"
Subject: Re: AMBER: solvatebox-solvateshell
Date: Mon, 8 Sep 2003 17:46:00 -0400

Hi

I am going to assume that what you call the solvate shell consists
in using a sphere to solvate your system instead of a cube as usual.
This representation actually saves computer time because you
use a sphere instead of cube (no box corners) and you don't have
images to take into account when calculating the interactions,
to say it simple. But you have to consider a bigger minimum
distance between your protein and the border of your sphere.
That is because of put extra forces on the solvent to avoid its
evaporation, they result in a bad description of the solvent at the
limits of the sphere. In addition there is the problem that David
mentioned that is due to the fact that your system is finite and thus
sees vacuum. People introduced some continuum representation
to decrease this effect. This representation is theoretically a
better representation of the natural system (no introduction of
periodicity) when the size of the sphere is big enough. One
additional problem is that you have to fix your system at center
of the sphere to make sure that it does not go play with your borders.

Anyway the time you would gain using the solvate shell is not
worth the trouble you are going to get to make people accept
your simulation. The periodic boundary conditions (PBC)
have been for a long time now the standard for molecular
dynamics simulation that is because they have proven to give
reasonable representation of the dynamics of proteins.

Nevertheless there are particular studies where the solvate
shell is useful:
1) FEP calculations
Allouche D, Parello J, Sanejouand YH
Ca2+/Mg2+ exchange in parvalbumin and other EF-hand proteins. A
theoretical study.
J Mol Biol. 1999 Jan 15;285(2):857-73.
2) Simulated annealing
R. Sankararamakrishnan, K. Konvicka, E.L. Mehler and H. Weinstein
Solvation in simulated annealing and high-temperature molecular dynamics
of proteins: A restrained water droplet model. Int. J. Quant. Chem.
(2000) 77: 174-186.

Hpe it'll help
XAvier