# Re: [AMBER] AMBER11 MMPBSA.py tutorial section 3.6

From: <Yokota_Akihiro.takeda.co.jp>
Date: Thu, 10 Feb 2011 15:04:56 +0900

Dear Jason,

As you mentioned, PB energies can be decomposed to use geometrical
arguments to partition the grid points.
But I'd like to know how to do it and how accuracy it is,
because it might bring us ambiguous and inaccurate energies.

And you mean that I must read sander source code rather than MMPBSA.py
because sander calculates PB energies which are decomposed into each residue.
---------------------------------------------------------------
I checked temporary file of MMPBSA.py, and found 2 files below.
This shows me that PB energies are calculated by sander.
Input file: _MMPBSA_pb_decomp_rec.mdin
Output file: _MMPBSA_receptor_pb.mdout
---------------------------------------------------------------

So, I will read sander source code, and if I can't understand it,
I will ask someone who developed sander module.

And I am grateful to you for creating MMPBSA.py which can recognize
receptor and ligand automatically, and make sophisticated operations.

Best Regards,
Akihiro Yokota

-----Original Message-----
From: Jason Swails [mailto:jason.swails.gmail.com]
Sent: Thursday, February 10, 2011 11:46 AM
To: AMBER Mailing List
Subject: Re: [AMBER] AMBER11 MMPBSA.py tutorial section 3.6

Hi Yokota,

As the GB equations consist of pairwise, analytical terms, you're right that
they're much easier to decompose than PB energies.

Since I haven't looked at the PBSA/decomp code, I can't say any of this for
sure. If I'm mistaken here, I'm hoping that a PBSA developer will step in
and correct me.

PBSA works by placing your system in a grid and solving the PB equation via
a finite difference method. It assigns each grid point a specific charge,
dielectric constant, and *starting* electrostatic potential (and ionic
strength if you specified one), which it then solves iteratively to get the
converged electrostatic potential. Thus, at the end of the calculation,
each gridpoint has associated with it a charge, potential, and dielectric.
You can use geometrical arguments to partition the grid points into
different residues (i.e. each grid point belongs to its *closest* residue).
Once you've created this partition, you simply toss it into the equation

DELTA G = 1/2 integral(charge density * reaction field potential)

Since it's a discrete problem, the integral becomes a sum, and for each
residue you just sum over that residue's grid points. I agree that it's not
as straightforward as GB, but it seems reasonable to me.

Hope this helps (and that it's right :) ),
Jason

On Mon, Feb 7, 2011 at 2:54 AM, <Yokota_Akihiro.takeda.co.jp> wrote:

> Dear AMBER users,
>
> I have a question about "Polar Solvation" energy for each residue in
> MMPBSA.py tutorial section 3.6.
>
> In the middle of the file below, there is a comment "Energy Decomposition
> Analysis (All units kcal/mol):
> Poisson Boltzmann solvent", I think this means that "Polar Solvation(PB)"
> of "each residue" can be calculated.
>
>
> However PB energy cannot be decomposed straightly by the definition (On the
> other hand, GB energies can be decomposed).
> Does anyone teach me how the PB energies are decomposed into each residue?
>
> Akihiro Yokota
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>

```--
Jason M. Swails
Quantum Theory Project,
University of Florida