[gmx-users] Iron containing structure simulation crash
jalemkul at vt.edu
Tue Aug 12 18:36:18 CEST 2014
On 8/12/14, 12:20 PM, Nikolaos Michelarakis wrote:
>> An output coordinate file is always produced unless mdrun completely fails
>> seg faults. That doesn't mean the coordinates are sane. It's meant to
> help you
>> diagnose in cases like this where you have an absurd force on some atom.
>> What is atom 2832? What's around it? Is there anything obviously
> clashing with
>> it? Also check the topology for errors.
> The atom is the iron that I introduced. It is bound to an oxygen which is
> itself bound to the carbon of a ring. It is found on the spot where the
> ligand, not introduced in the simulation yet, is binding to the protein.
> There are a few carbons and nitrogens but nothing in close proximity or
> clashing with it.
> The topology seems ok to me. This is the link to the file, the iron shows a
> charge of +3, could this be the cause of such a great force?
Given that that's a really unphysical situation, it's certainly a contributing
factor. As I'm sure I've said before, additive models fail spectacularly for
metals like these. That is a massive charge density, when in reality, the
charge is spread over the ligating molecules. You'll probably have to do some
real parametrization to develop suitable charges and LJ terms, rather than
relying on whatever there might already be.
In isolation, the topology doesn't help much. A full force field directory and
the coordinate file would illuminate any obvious errors.
Justin A. Lemkul, Ph.D.
Ruth L. Kirschstein NRSA Postdoctoral Fellow
Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 601
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201
jalemkul at outerbanks.umaryland.edu | (410) 706-7441
More information about the gromacs.org_gmx-users