AMBER Archive (2002)

Subject: Re: BLOWUP after 100 ns???

From: David A. Case (case_at_scripps.edu)
Date: Thu Sep 19 2002 - 19:29:58 CDT

On Thu, Sep 19, 2002, Randy M. Wadkins wrote:

> Under the general heading of "What the...??" After over 50000 MD
> steps, I get the following, using the following parameters in sander,
> from Amber 7:
>
> run MD at 300K
> &cntrl
> imin=0,
> ntc=2, ntf=2,
> cut=12.0, igb=1, saltcon=0.2, gbsa=1,
> ntpr=50, ntwx=50, ntwe=50,
> nstlim=500000, dt=0.002,
> ntt=1, tempi=300.0, temp0=300.0, tautp=2.0,
> ntx=5, irest=1, ntb=0, ntwr=500,
> nscm=500,
> &end
>
> Note that this is using the generalized born solvation, no periodic
> box, and no explicit solvent, no constant pressure. It's also a 1
> residue mutant of a protein that the simulation completed without a
> hitch.
>

(Note that this is a blowup after about 100ps, not 100 ns...)

There was a similar report on the reflector a short time ago. I'm a little
suspicious of using the Berendsen temp. algorithm for a non-condensed phase
system with a relatively small number of degress of freedom: one could get
hot-spots in the molecule (I imagine) that might not be properly handled. But
this is just a guess. We have run temperature regulated GB runs on smallish
proteins (100 residues) for 10-20 ns without seeing behavior like this (yet).
We do know that the igb=1 formula can lead very occasionally to excessively
large effective radii, which can in turn lead to bad dynamics.

It's also worth noting that the "sa" part of gbsa (as implemented in Amber)
has some known "discontinuities" that disrupt energy conservation; see

%A V. Tsui
%A D.A. Case
%T Theory and applications of the generalized Born solvation model in
macromolecular simulations
%J Biopolymers (Nucl. Acid. Sci.)
%V 56
%P 275-291
%D 2001

And of course, a cutoff of 12 Ang. will also introduce noise.

Some things to try:

(1) you may need to set vlimit lower than the default. Try 10. Maybe
     even 8. or 5....
(2) it may help to go to 0.001 for dt
(3) you could set gbsa=0, ntt=0, cut =15. Of course, this looses the SA
     part, but should conserve energy reasonably well (after everything
     is equilibrated) without the need for temperature regulation.

I'd try (1)+(2) myself to see if that made a difference.

..good luck...dac 

-- 

==================================================================
David A. Case                     |  e-mail:      case_at_scripps.edu
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            |    http://www.scripps.edu/case
==================================================================