Re: Internal dielectric in MM-GBSA

From: David A. Case <>
Date: Mon 10 Feb 2003 12:29:44 -0800

On Mon, Feb 10, 2003, Robert J. Woods wrote:
> To more reasonably approximate the interior of a protein, we are interested
> in using a value of the interior dielectric greater than 1, perhaps 3 or 4.
> There are two variables that affect this in GB, namely, either by setting
> dielc=n OR intdiel=n (done using a modified script)
> Both give the expected results for <Eelec> that is, they scale it by 1/n.
> But, they behave differently toward <GB>. Dielc=n scales <GB> by 1/n, but
> intdiel scales by a little more. For example, (to keep the math clear) if I
> set n=10 here is what I get:
> dielc=1.0 (default) <Eelec> = 54.6, <GB> = -14.0
> dielc=10.0 <Eelec> = 5.46, <GB> = -1.40
> intdiel=10.0 <Eelec> = 5.46, <GB> = -1.25 <------
> Shouldn't dielc and intdiel both scale <GB> by 1/n ?

sigh...basically, you should not use "dielc" with GB. The manual is
probably not quite explicit enough. It says "this [dielc] is NOT the
solvent dielectric for GB", but I guess it also needs to say also that this
parameter is not to be used with GB; the code should probably make
this combination illegal as well, to reduce confusion.

Basically, "dielc" was put in the code twenty years ago...all it does
is mutliply all of the charges by 1/sqrt(dielc). This would be correct
for a uniform dielectric, (in your example, if the dielectric inside *and*
outside the protein were 10); it is not the correct procedure for a mixed
dielectric system, such as that GB is trying to model.

For GB, you need to set the values "intdiel" and "extdiel" to the values
you want. I will emphasize what the Users' Manual says about "intdiel":
"Default is 1.0. Other values have not been extensively tested." You should
take this to mean that you can see if values of 3 or 4 (or whatever) give
useful results, but you are perhaps just getting enough rope to hang yourself.
You are essentially experimenting with a new energy function.

..good luck....dac

David A. Case                     |  e-mail:
Dept. of Molecular Biology, TPC15 |  fax:          +1-858-784-8896
The Scripps Research Institute    |  phone:        +1-858-784-9768
10550 N. Torrey Pines Rd.         |  home page:                   
La Jolla CA 92037  USA            |
Received on Mon Feb 10 2003 - 12:29:44 PST
Custom Search