Re: [AMBER] excluding atom sets from MMPB/GB-SA calculations

From: Jason Swails <jason.swails.gmail.com>
Date: Wed, 17 Dec 2014 22:35:48 -0500

> On Dec 17, 2014, at 3:33 PM, Hallel Freedman <hfreedma.ualberta.ca> wrote:
>
> Hi Jason,
> Yes actually I put a large restraint (100kcal/mol) on the phosphate groups
> during my simulations, which I am hoping should help in the cancellation of
> error terms. If what I am suggesting is not a problem with the
> implementation itself, then could you please point me to how to most easily
> to modify the code for this?

I don’t see this as something you can implement with PB or GB by changing the code that you can’t do by a simple modification of the topology file.

> My concern with stripping the phosphate atoms or changing their charges to
> zero is that this would effectively place either a cavity or a hydrophobic
> group within a very hydrophilic receptor pocket which may also introduce
> large, this time positive, error terms to the binding free energies -
> wouldn't you agree?

There’s going to be an effect of just throwing away the phosphate no matter what you do. Since the phosphate stabilizes binding, getting rid of said phosphate will clearly destabilize the binding. I had thought that your argument was that this change was basically the same for every ligand and would therefore cancel.

As far as the PB equation goes -- the phosphate is either there or it’s not. It either affects the dielectric boundary or it doesn’t. If you strip the phosphate atoms out, then you introduce what the PB equation will treat as a solvent-filled cavity. If you keep the phosphate atoms in but change the charges to zero, you introduce what is effectively a cavity with a low dielectric. If you want to get rid of the phosphate, those are your two choices: there is either a pocket of high or low dielectric where the phosphate *should* be.

If you want the region to have a high dielectric, use ParmEd or cpptraj to strip the phosphate group. If you want the region to have a low dielectric, keep the phosphate atoms there and change their charges to zero. I don’t see either way as being more correct than the other. I am skeptical of both approaches, so if someone like me was reviewing this approach for publication you would need to make strong arguments to convince me.

I would probably recommend a more rigorous approach to solving the electrostatics WITH the phosphates there -- perhaps solving the non-linear PB equation or using MM/3D-RISM instead -- before trying to hack away the phosphate.

HTH,
Jason

P.S. -- Another thought. There are many times when we devise methods that do things that are physically impossible -- think alchemical free energy methods like TI or FEP where you simply “grow” a particle into existence by jumping between states where the particle *partially* exists. The way we justify this is by recognizing the free energy is a state function that is path-independent, and we construct a thermodynamic cycle to base its use in rigorous theory.

Try doing the same here -- try to construct some thermodynamic cycle whereby you can compute some energies without the phosphate. The legs of that thermodynamic cycle will tell you exactly what you need to compute and should shed light on the types of approximations and assumptions you are making to do so.

--
Jason M. Swails
BioMaPS,
Rutgers University
Postdoctoral Researcher
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Dec 17 2014 - 20:00:02 PST
Custom Search