Re: [AMBER] free energy of solvation of small molecules

From: Hannes Loeffler <>
Date: Thu, 3 Dec 2015 11:46:45 +0000

On Wed, 2 Dec 2015 16:49:20 +0200
Fabian gmail <> wrote:

> Dear Jason,
> Thanks a lot for the detailed explanation, so in the paper I am
> following, they minimised once, and then use the same minimised
> molecule (in this case a drug Improving the Efficiency of Free Energy
> Calculations in the Amber Molecular Dynamics Package Joseph W.
> Kaus,*,† Levi T. Pierce,†,‡ Ross C. Walker,†,‡ and J. Andrew
> cCammon†,§,∥,⊥) and then perform IT calculation.
> So the protocol I will be using will something like this:
> 0) minimize 20,000 steps ONCE for lambda = 0. 5 and then for each
> lambda for lambda 0, 0,1, 0,2, etc. … 1 do the following:

That's just overkill. All you want to do with the minimisation step is
to eliminate close contacts which could lead to high forces in MD and
thus potential instability. A few hundred steps should be enough.

You could do the same protocol for every lambda as it costs very
little. Keep in mind that you are extrapolating away from the
references state and thus "equilibration" of later lambdas
potentially needs more time. Not sure what the convergence behaviour
for decoupling is but at the end-point your molecule will have created
a void and will be filled with water.

You could take over the data from the previous neighbour lambda if you
wanted to but in practice this means that you would have to run all
lambdas one after another.

> 1) heating to 300 K during 500 ps at constant pressure
> 2) equilibrate at constant volume during 500 ps
> 3) production for 5 ns at constant volume x 3
> This protocol is different than in the A9 tutorial, but sounds more
> right for my purpose, does it sounds a right protocol? This is to
> calculate the DG of solvation (disspereance of a drug in water).

With constant volume you will get the Helmholtz free energy, not Gibbs.


AMBER mailing list
Received on Thu Dec 03 2015 - 04:00:06 PST
Custom Search