Subject: Re: AMBER: Calculating solvation energies for MM-PBSA use
From: samantha hughes
Date: 18 Sep 2003 14:13:04 +0100

Hi Thomas,

See this publication where we describe our MgATP MM-PBSA calculations:

S.J. Hughes et al, BMC Structural Biology 2003 3:5

its available under openaccess at:
http://www.biomedcentral.com/1472-6807/3/5

Even though your protein is big, I'd suggest using all the coordinates
for MM-PBSA calculations. i am sure this will influence your results,
especially the Delphi PB calculations. Also, your choice of internal and
external dielectrics will strongly affect your results! If the delphi
calculations are really expensive on a large system, run it over just a
few snapshots (ca 10) to get a feel for the results using that
dielectric constant.

Good luck!

Samantha


On Thu, 2003-09-18 at 13:43, Thomas Steinbrecher wrote:
> Dear AMBER users,
>
> I have some difficulties in using MM_PBSA to calculate the
> binding free energy of MgATP bound to a protein binding
> pocket. I use AMBER7 on Linux.
>
> I have set up my system in xleap using the protein pdb-file
> and the nucleotide parameters of Dr. Carlson from the AMBER
> homepage. I equilibrated my system for 200ps and I have
> analysed the following nanosecond of dynamics. Since the
> protein is rather big, I have defined a shell of 25A around
> my ligand and allowed only those residues to move using the
> ibelly- command, similar to the calculations in the
> biotin/streptavidin tutorial on the AMBER homepage.
> Solvation was simulated by adding a Watercap of 25A radius
> centered on the ligand and including these waters in the
> moveable residues.
>
> For the MM-PBSA calculations I wrote the coordinates of
> amino acid residues within 20A of the ligand into snapshots
> and ran the MM-PBSA-script over them. I know that by using
> only a cutout of the protein I introduce an incorrect
> protein-water-contact and a wrong dielectric constant 20A
> away from my ligand. I assume that leads to the same error
> in the complex and receptor snapshots, so the energy
> difference should nevertheless be meaningful. I use delphi
> V4 release1.1 to calculate the PB part.
>
> The result, however, is a total binding energy of more than
> +100 kcal/mole (Without entropy contribution, which would
> make the value even more positive). Since the gas phase
> energy is about -500 kcal/mole I presume something is wrong
> with my calculation of the solvation energy.
>
> Has someone experienced similar problems in MM-PBSA
> calculations with Magnesium-nucleotides?
> Could there be anything wrong with my setup?
>
> Many thanks in advance for any helpful comments and sorry
> (again) for the long post,
>
> Kind regards,
>
> Thomas Steinbrecher