Re: AMBER: free energy profile calculation

From: David Smith <>
Date: 15 Jan 2004 15:37:27 +0100

Hi Martina,

I will attempt to give you my two cents worth but wish to add that it
may only be worth about half that value in actuality.


As I read the manual and the code, the constraints section of gibbs is
designed to handle atomic number specifications of constraint
definitions and not COMs. Therefore, from the start you would need to
either (a) write some code to allow the COM to be used or (b) abandon
the idea and use, for example, an atom of the molecule that approximates
the the COM. If your molecule is semi rigid (SHAKE) and you don't expect
wild dihedral fluctuations, I would guess there wouldn't be a whole lot
of difference in the two approaches.

However, this only takes care of one end of the constraint. In my
opinion, that is the easy end. Again, as I read it you are unable to
constrain just the z-coordinate of your "COM atom" but you need to apply
a distance constraint with another end. Apart from the fact that you
would lose full mobility in the x and y directions with this approach,
somehow you have to come up with a definition of the other end of the
constraint (or restraint). Unfortunately, I don't see a nice way to do
this. If you choose a particular solvent molecule, it is possible that
it would be pulled out of the solution (depending on various factors)
rather than your gas phase molecule being pulled in. Ditto if you try to
define some dummy atom or vdw particle.

Even if you manage to get past that obstacle, you are still going to run
into several problems using EWALD in gibbs. That is, it is possible in
single processor mode and possibly using the old shared memory SGI
parallel mode. However, it won't work in parallel with any mpi
implementation. At least it is not supported by the AMBER team and my
coding skills are not up to getting it working.

So, in summary, if you are brave you could try:

Defining a CONSTRAINT between a "COM atom" and an oxygen atom of a
particular water and run it on a single processor. This would severely
limit the xy mobility of the "solute" and if you use the potential
forces method, the contributions of the chosen water would be included
in the free energy. You also have to hope that the water stays in
solution (use at least IBIGM=0).

You could also try umbrella sampling in GIBBS (basically the same as
above but you would have to play with the potential a bit) or in SANDER.

That is, there may be a way to define a more flexible kind of restraint
force in SANDER and calculate the PMF using some kind of histogramming
technique. However, this is a bit beyond my area of expertise. Perhaps
someone else would care to comment on this option.

I hope that is of some kind of minimal use.


On Wed, 2004-01-14 at 22:14, Martina Roeselova wrote:
> Hi,
> I am looking at interactions of small gas phase molecules with the
> surface of various liquids (water and solutions of different salts). To
> simulate the vacuum/liquid interface, I use a slab geometry employing
> periodic boundary conditions, and I NEED to use Ewald for my systems. I
> use SHAKE to freeze the intramolecular degrees of freedom in both the
> water molecules as well as my gas phase molecule.
> I would like to calculate the Gibbs free energy profile for a gas phase
> molecule approaching the liquid surface along the normal to the
> vacuum/liquid interface (z-axis).
> |z
> vacuum |
> |
> -------------------------------------
> liquid
> -------------------------------------
> vacuum
> I want to perform a series of runs in which I constrain ONLY the
> z-coordinate of the CENTER OF MASS of my gas phase molecule within
> certain window along the z-axis, allowing the gas phase molecule to move
> in the x-y plane as well as rotate freely during each run (the
> intramolecular vibrations of the molecule frozen using SHAKE). Is there
> a possibility in AMBER to constrain the center of mass of a selected
> group of atoms in such a way? Can someone teach me how to do it?
> Thanks
> Martina
> -----------------------------------------------------------------------
> The AMBER Mail Reflector
> To post, send mail to
> To unsubscribe, send "unsubscribe amber" to
Dr. David Smith
Department of Chemistry
Ludwig Maximilians University
Butenandt-Str. 5-13, D-81377 Munich
Tel.: +49 (0)89 2180 77740
Fax.: +49 (0)89 2180 77738
The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to
Received on Sun Jan 18 2004 - 19:53:02 PST
Custom Search