Re: AMBER: solvateOct method

From: David A. Case <>
Date: Fri, 25 Feb 2005 08:53:19 -0800

On Wed, Feb 23, 2005, Cameron Mura wrote:
> I'm wondering if anyone could explain (briefly or in detail!) the method
> used by LEaP's "solvateOct" functionality for setting-up truncated
> octahedral PBCs? In particular, I'm wondering about the transformations
> applied to the solute and how the corners of the cube are shaved off (or
> whatever algorithm is used to construct it)??
> The reason I ask this is that I get vastly different results in terms of
> number of TIP3 waters depending on whether I (a) go from an initial PDB
> file to a solvated octahedral box in xleap, using the solvateOct command
> w/ a buffer of, say, 10 A, versus (b) set-up and solvate the system
> using other software (like VMD and its psfgen utilities) to immerse the
> solute in a large cube and then lop off the corners. The difference is
> quite large (~13,000 waters from scenario (a) vs ~19,000 from (b)), and
> what really peaked my curiosity is that I get a similar discrepancy in
> num of waters if I manually construct the box (i.e., route (b)) using
> the solute after it's already been reoriented by xleap's solvateOct
> function.

Tom Cheatham or Mike Crowley should really answer this, but I will try.

The solvateOct command actually sets up an equivalent triclinc box, which can
then be transformed into a truncated octahedron. To see this, put your
coordinates into ptraj, and execute the "image" command with and without the
"familiar" keyword. Then save the output and visualize in xleap or in your
favorite viewing program. With "familiar" you will get something that looks
like a truncated octahedron; but inside Amber, the system is really the one
you will see without the "familiar" keyword.

So, you are correct: this procedure is not equivalent to creating a cube and
lopping off the corners. The following paper has a discussion of some of
these points:

%A H. Bekker
%T Unification of Box Shapes in Molecular Simulations.
%J J. Comput. Chem.
%V 18
%P 1930-1942
%D 1997

As I said earlier, others actually know more of the details here....

The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to
Received on Fri Feb 25 2005 - 17:53:01 PST
Custom Search