AMBER Archive (2009)

Subject: Re: [AMBER] Compiling and running NAB programs in parallel using MPI

From: case (
Date: Tue Jun 30 2009 - 11:02:57 CDT

On Tue, Jun 30, 2009, Kaushik Raha wrote:
> I am very interested in using the LMOD package for conformational search of
> protein-ligand complexes. But currently it seems like these calculations are
> very expensive. LMOD seems to be using xmin for the minimization part so I
> was interested in its parallelization. Do you have any other suggestions for
> speeding up these calculations besides parallelization?

The next version of AmberTools will allow part of the system to be frozen,
which can greatly speed up the calculation if you only need to optimize the
parts of the structure around the ligand binding site. But it is certainly
true that lmod calculations are typically rather expensive.

> For example, I have noticed in the lmod example of beta_secretase, the
> gb cutoffs are very liberal (cut=99, rgbmax=99). Shorter cutoffs would
> obviously speed up the calculations so I was wondering if there is a
> physical reason behind these cutoffs w.r.t an LMOD calculation, in the
> example?

You should be able to use cutoffs in the range of 15-20 Ang. Maybe Istvan or
others will have comments here.

> I also came across some problems with running a gbsa (gbsa=1 to include
> the non-polar part) calculation on protein-ligand complexes. The problem
> appears to be in LCPO routine for the non-standard ligand.

It's hard to help without more information. Does the calculation (say, just a
minimization) work in sander? Are there any error messages? etc.


AMBER mailing list