[MMTK] Normal Modes for protein complexes

Konrad Hinsen hinsen@cnrs-orleans.fr
27 Feb 2002 11:48:20 +0100

"Ben Carrington" <bjc32@cam.ac.uk> writes:

> Can MMTK handle normal modes calculations for protein complexes that
> are 5500 atoms and above? Hardware is not a problem, but if a
> particular configuration works best please let me know.

If you have the memory (2 GB for the Hessian matrix alone), it should
work (assuming that the eigenvalue routines in LAPACK can handle it, I
never tried such a large system because I don't have the memory).

However, before launching a brute-force normal mode calculation for a
large system, look at all the special options that MMTK has for normal
modes. You could use a C-alpha model with the Deformation force field,
that would leave you with something like 400 atoms, i.e. a few-minute
calculation on a standard PC. For the low frequency modes, you'd get
the same information.

If you do want to work with an all-atom model, but need only
low-frequency modes, you could try subspace normal modes with the
Fourier space. Finally, if you want the high-frequency modes, just cut
your molecule into pieces and study them separately.

The biggest protein complex I ever treated with MMTK had 8700
residues, I used a C-alpha model plus a Fourier subspace.

Konrad Hinsen                            | E-Mail: hinsen@cnrs-orleans.fr
Centre de Biophysique Moleculaire (CNRS) | Tel.: +33-
Rue Charles Sadron                       | Fax:  +33-
45071 Orleans Cedex 2                    | Deutsch/Esperanto/English/
France                                   | Nederlands/Francais