[MMTK] MMTK beguinner

[khinsen at cea.fr]

On Apr 19, 2005, at 11:39, Matias Saavedra wrote:

> I would like to know which is the better method to determine the 
> amplitude of the mouvements of selected residus, for example from the 
> active site, DEFORMATION ENERGY or FLUCTUATIONS() from NormalModes 
> class.

Those two quantities measure different things. Fluctuations measure how 
much a residue moves around in space, i.e. how much its absolute 
position varies. Deformation energies measure how much a residue moves 
relative to its geometrical neighbours in the protein.

Many people confuse the two and consider fluctuations a measure for 
local flexibility in a protein, the literature is full of such 
references. But repetition doesn't make it true. To illustrate the 
difference, consider a big protein with two domains that move relative 
to each other. The deformation energy will then be high in the hinge 
region between the two domains, but the fluctuations will be highest at 
the points in the domains that are far away from the hinge, whereas the 
hinge region will have fluctuations close to zero.

> In this context i used the method fluctuations() to calculate the 
> mouvements of residues in a given protein,  with the  simplified model 
> using deformation force field and i obtained very low values (around 
> 0,00005).

Absolute amplitude values from normal mode calculations are never 
reliable at the current level of force field development. The 
deformation force field in MMTK includes an arbitray scaling factor, so 
the amplitudes are meaningless by design. However, you can use 
amplitude information from experiments to establish an amplitude scale 
for your particular protein. For details, see

    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)

However, you should better use the CalphaForceField for such 
applications, which is the one described in that paper.

> I would like to know the units of the fluctuations results and if the 
> values i get are in the correct range.

The units are nm^2, but given the arbitrary scale factor, that doesn't 
help in your case.

> Finally i would like to know how many modes i have to consider to make 
> a good fluctuation analysis.

That depends on your application and the size of your protein. If using 
twice as many modes yields a significantly different result, you didn't 
use enough. As a rule of thumb, start with 5% of the total number of 
modes, then check with smaller and larger numbers.

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
---------------------------------------------------------------------