[MMTK] Energy calculations

Konrad Hinsen hinsen at cnrs-orleans.fr
Mon May 17 12:43:44 CEST 2004

On May 14, 2004, at 17:38, Shirley Hui wrote:

> extremely negative compared to energy values I get from other 
> applications.
> For example, I have compared the energies obtained from MMTK (amber 
> 94) to
> applications like MOE and TINKER (charmm27).  A summary of the results 
> is

Without knowing the details of the calculations you made, it is hard to 
say something about them. Absolute energies don't mean much anyway, 
energies are defined only up to an additive constant. Moreover, small 
configurational changes can cause big energy differences, meaning that 
a configuration that is reasonable for one force field can be very bad 
for another one.

You just give PDB codes, but you don't say how you constructed the 
internal models from these files and for what exactly you calculated 
the energies. Note that most PDB files lack hydrogen atoms, some 
programs add them automatically (and assign them positions according to 
different algorithms), some don't. Moreoever, a configuration straight 
from the PDB is likely to be energetically bad with most force fields.

If you want to do a meaningful comparison between force fields, you 
need to do some sampling (e.g. by MD) for all of them and compare e.g. 
energy fluctuations. Better yet, try to calculate free energy 
differences (which is a lot of effort, unfortunately). With 
single-configurations comparisons, all you can test is two different 
implementations of the same force field.
Konrad Hinsen
Laboratoire Léon Brillouin, CEA Saclay,
91191 Gif-sur-Yvette Cedex, France
Tel.: +33-1 69 08 79 25
Fax: +33-1 69 08 82 61
E-Mail: hinsen at llb.saclay.cea.fr

More information about the mmtk mailing list