[MMTK] Re: MMTK beguinner

khinsen at cea.fr khinsen at cea.fr
Tue Apr 19 17:13:32 CEST 2005

On Apr 19, 2005, at 15:29, Matias Saavedra wrote:

> Ok i get the idea. So i think that deformation energy is the usefull 
> magnitude for the analysis i want to carry out. I want to determine 
> the flexibility in the active site of several proteins.

Yes, go for deformation energy in that case.

> The problem of the scaling factor for the results of fluctuations 
> applies to energy deformations as well? Or the energy deformations 
> magnitudes calculated with the CalphaForceField can be used directly?

The problem of the scaling factor exists in all methods, as it is quite 
fundamental. There is very little experimental information on the 
amplitudes of large-scale motion, and none of it was ever used in force 
field design. So even with detailed force fields and detailed MD 
simulations, you have no guarantee whatsoever to obtain reasonable 
information about the amplitudes.

With normal modes, most people study local minima of all-atom 
forcefields such as Amber. It is well known, and easy to understand, 
that amplitudes are grossly underestimated in this way. With all other 
approaches I am aware of, including my own of course, the global 
amplitude is defined by a factor that needs to be fitted to 
experimental data in order to obtain reasonable amplitudes.

However, for many calculations the absolute amplitudes are not 
required. If you use the CalphaForceField out of the box, the relative 
amplitudes of fluctuations or deformation energies for different 
residues will be quite good.

> Another question:
> The simplification of the system by considering only Calpha carbons is 
> sensible enough to detect changes in flexibility produced by point 
> (single) mutations? or should i use for that purpose the complete 
> Amber94 analysis?

The sidechains don't enter at all in a C-alpha model, so if you take 
the same atomic structure and exchange sidechains, there will be no 
change at all. If however your mutation leads to a different structure, 
then the structural change will lead to a change in flexibilities.

Unless the mutation causes clear structural changes, I don't think any 
normal mode technique will be of use. The dynamics in the local 
environment of a sidechain consist of transitions between rotamers, 
methyl group rotation, sidechain deformation, and similar motions. None 
of them is well approximated by a harmonic potential (actually, this is 
an understatement). Better go for MD, and use time correlation 
functions to characterize the dynamics.

If however you are interested in the kind of large-amplitude motions 
that normal modes describe well, then you can safely assume that a 
single-point mutation won't make any difference, unless there is a 
significant change in structure.

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