[gmx-users] changing atom types versus adding dihedrals to atom types

Justin Lemkul jalemkul at vt.edu
Wed Sep 18 02:40:59 CEST 2013

On 9/17/13 8:18 PM, Rafael I. Silverman y de la Vega wrote:
> Can you give some examples of how these verifications are  different for
> different force fields? It doesnt seem like verifying takes that much time,
> but a theorist prof in my department told me not to worry as long as my
> system doesnt blow up...

IMHO simply not blowing up tells you nothing.  I can show you a dozen 
simulations that don't blow up that have terrible small molecule topologies that 
produce bad results.

Parametrization methods and validation procedures are defined in the literature 
and one could easily fill a book chapter (or more) on such topics, so I will not 
go into it in an email.  You may have to go back several years (or even decades) 
in the literature to get the full story.

> And what do you mean "thourough parametrization?

Most people hope for a simple, one-shot step they can take to parametrize a 
small molecule.  There are numerous "black box" methods out there, some good and 
some bad.  I advise people to be thorough in terms of what the force field 
requires and what their chemical knowledge tells them.  For instance, for water 
interactions in CHARMM, HF/6-31G* works well for most compounds, unless sulfur 
is involved, in which case we need to do a more expensive MP2/6-31G* 
calculation.  You can get an OK result for everything with HF, but it's not 
sufficiently accurate in all cases.

> I parametrized flavin mononucleotide using amber99sb-ildn, I used existing
> atomtypes in the force field, but I added partial atomic charges based on a
> decent DFT calculation in orca, and I had to add 2 distance restraints on
> the delta negatively charged phosphate oxygens to keep them from crashing
> into the delta positive hydrogen on the same phosphate. Is that thorough in
> your opinion?

How does it compare with the results of running the molecule through 
antechamber?  Usually GAFF gives a reasonable topology with minimal adjustment 
necessary.  That's one of the benefits of Amber; there are very well-defined 
protocols and a robust general force field for the parametrization.



Justin A. Lemkul, Ph.D.
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 mailing list