Re: [AMBER] minimization segmentation fault

From: Bill Ross <ross.cgl.ucsf.EDU>
Date: Fri, 30 Jul 2010 09:52:50 -0700

What is the type of the hydrogen? What vdw params does that
type have? It's unlikely that it is trying to merge with another
hydrogen, so if there's a way to print out crds for each step
of min, it would be interesting to watch the movie in VMD and see
what is happening. Failing that, you could set the maxcyc to
one step less than you see blowup, so that you'd have a restrt
from right before the blowup to look at (i.e. make pdb). The
movie will be more instructive.

It might be interesting to run a few hundred steps of 10K MD
to see if it loosens things up.

Bill

> So I get some very high Gmax value during minimization for a hydrogen
> atom(HG2) bound to a Arginine - when I look at the pdb file the distance
> between its nearest neighbor is 1.76 to a hydrogen on proline residue

> I use constraints for the hydrogen bond length through shake
> SHAKE:
> ntc = 2, jfastw = 0
> tol = 0.00001

> the rms gradient is 5.8 which is pretty high. The vdwaals are negative
> -7764. I have not yet added any extra water molecules only the crystal ones
> not having too high clashing. Are there any debugging in order to identify
> the clash - is it vdw or something else and how to resolve that?

> On Tue, Jul 27, 2010 at 8:53 PM, Bill Ross <ross.cgl.ucsf.edu> wrote:

> > > Furthermore, the residues around are displaced which
> > > they are not supposed to do - I have included low vdw radii and the
> > bonding
> > > energy has a very high force constant (100 ) - can I get the clashing
> > part
> > > of Gmax out?
> >
> > The atom indicated by GMAX is the one to focus on.
> > What is it? What type? What vdw params? What bonded to?
> > Are you expecting that a high bonding force constant will somehow
> > compensate for low vdw radii? That would only work in a highly
> > constrained system, likely with just a few atoms that could not
> > form vdw contacts.
> >
> > > When I look in the pdb it seems fine so there is a clash I am
> > > not able to see.
> >
> > The clash probably develops as atoms are pulled together.
> > If you print crds for every step of md or min, you likely will
> > see the problem developing.
> >
> > Bill
> >


--===============0621702753713782685==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber

--===============0621702753713782685==--
Received on Fri Jul 30 2010 - 10:00:04 PDT
Custom Search