[MMTK] Normal Modes problem

khinsen at cea.fr khinsen at cea.fr
Wed Jun 1 18:47:31 CEST 2005


On Jun 1, 2005, at 18:28, Vasisht R. Tadigotla wrote:

> I'm using the sparse matrix with the subspace modes and 
> CalphaForceField,
> and it generates the low frequency normal modes in ~10 min. Is there 
> any
> advantage in doing the full normal mode calculation with the sparse 
> force
> constant matrix, I'm only interested in the domain motions ? Would it 
> make

A full normal mode calculation yields a (3*3300)^2 matrix - you risk 
putting your machine in swapping mode again.

In terms of results, the larger the subset the more accurate the modes, 
but for most applications the difference is small. You will likely get 
the same domains and a very similar B-factor profile from a full 
calculation. The easiest way to get an idea of the differences is to do 
both calculations on a smaller protein.

> a large difference in terms of the calculated B-factors ? I would also
> like to estimate the frequency of motion.

The deformation force field does not give absolute frequency values, 
nor absolute B factors. There is a global scaling factor in the model 
that has to be fitted (to experimental data) for each protein if 
absolute values are required.

Note that there is no theoretical model at the moment that does not 
have this problem. Standard normal modes (in a local minimum) yield 
fluctuation amplitudes that are systematically too small, and 
frequencies that are much too large compared to the real time scales of 
the diffusional motion that characterizes slow protein dynamics. 
Quasiharmonic modes don't get decent amplitudes either because MD 
trajectories are always too short to provide good sampling of slow 
motions.

For a discussion of these issues, I recommend the following paper:

     K. Hinsen, A.J. Petrescu, S. Dellerue, M.C. Bellissent-Funel, G.R. 
Kneller
     Harmonicity in slow protein dynamics
     Chem. Phys. 261, 25-37 (2000)

Konrad.
--
---------------------------------------------------------------------
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: khinsen at cea.fr
---------------------------------------------------------------------




More information about the mmtk mailing list