Re: [AMBER] Fast ibelly?

From: David Minh <>
Date: Tue, 20 Mar 2012 14:10:15 -0400

Hi Ross,

Thanks for the reply.

What I meant by fast is skipping all the belly-belly interactions, and also using an implicit solvent model (then PME issues are bypassed). I don't think there is a fundamental reason that belly shouldn't work with igb>0 because something similar works in NAMD. NAMD has a feature to fix atoms, which allows for a very fast simulation of a fixed protein and flexible ligand. Unfortunately, it seems like their implementation of OBC generalized Born doesn't take advantage of fixed atoms, as it is only marginally faster than a fully flexible implicit solvent simulation. I don't know if it would be easier to accelerate the GB implementation in NAMD or belly calculations in AMBER, but AMBER has a wider variety of and greater flexibility with implicit solvents, including the critical ability to separate cavity and van der Waals terms in the nonpolar solvation free energy.


On Mar 20, 2012, at 1:09 PM, Ross Walker wrote:

> Hi David,
>> Would it be difficult to make "belly-type" dynamics fast? I'd like to
>> see whether using AMBER with a fixed protein and flexible ligand can do
>> molecular docking well. (As I understand it, the current use of AMBER
> It depends what you mean by 'fast'. For PME runs you still have to do the
> full FFTs and you also have to do non-belly to belly atoms, so you only save
> by skipping the belly-belly interactions which in the grand scheme of things
> is not generally that much saving, especially considering the size of the
> additional approximation you are adding. It would probably be better to look
> at ways of calculating some fixed charge field and applying that to the
> underlying equation but I haven't looked at it much.
>> with DOCK is as a postprocessing step in which all the molecules are
>> able to move, even if there are harmonic restraints, and this isn't
>> faster than a normal AMBER run.)
> It is a little bit faster, just not significant. It could possibly be
> improved by adding logic to truly skip all belly-belly interactions, bonds,
> angles etc etc. I think at the moment that is not done, the belly atom
> forces are just zeroed.
>> Also, why is ibelly unavailable with igb>0?
> This is a tricky one. I attempted to code it a long time ago. You can apply
> belly, I believe, for atoms 1 to N but not for say 100 to N. I.e. it has to
> be continuous and start from 1. I attempted to add a feature to the code to
> allow arbitrary choices with GB but it never really worked properly and not
> enough people use Belly to make it worthwhile figuring what was wrong. I am
> not sure if it was a coding issue or something more fundamental. I believe
> the problem might stem from an incorrect consideration of the forces when
> doing belly. There is a force due to the GB radii which are calculated
> including the belly atoms. This is clearly not correct and so one needs to
> figure out how to deal with this but I think the solution is non trivial.
> You can't simply zero the forces since there are components there from the
> GB radii which are not being 'bellied' as it were.
> All the best
> Ross
> /\
> \/
> |\oss Walker
> ---------------------------------------------------------
> | Assistant Research Professor |
> | San Diego Supercomputer Center |
> | Adjunct Assistant Professor |
> | Dept. of Chemistry and Biochemistry |
> | University of California San Diego |
> | NVIDIA Fellow |
> | | |
> | Tel: +1 858 822 0854 | EMail:- |
> ---------------------------------------------------------
> Note: Electronic Mail is not secure, has no guarantee of delivery, may not
> be read every day, and should not be used for urgent or sensitive issues.
> _______________________________________________
> AMBER mailing list

AMBER mailing list
Received on Tue Mar 20 2012 - 11:30:05 PDT
Custom Search