<<<< back

 Free energy calculation

General theory. There is a variety of methods for computing the free energy difference between given equilibrium states of a molecule or complexes of molecules (see e.g. "Computer Simulations of Biomolecular Systems", Vol 1 and 2, W. F. van Gunsteren et al. , Escom, 1989, 1993). These methods refer to transitions that are induced by changing the Hamiltonian. When we started to simulate conformational changes along constrained reaction coordinates of the distance type it turned out that no practicable method was available to compute the corresponding free energy changes. 

Reaction coordinates are a useful means for inducing chemical reaction and conformational changes. Note that a constraint imposed to such a coordinate leaves most degrees of freedom really free (for a protein in solution 10 – 100 thousands). For TMD and XTMD we use reaction coordinates of the distance type like the rms distance between conformers or the radius of gyration. Our 2001 paper is a survey of the particular properties of such coordinates which make them a useful tool for the simulation and thermodynamic evaluation of conformational changes.

Free energy of a reaction coordinate. In the early nineties there was no practicable method for computing free energy profiles along an arbitrary reaction coordinate (rc). The formulation along the lines of thermodynamic integration would require definition of a complete new set of coordinates including the rc, and computation of all corresponding partial derivatives as stated most clearly by den Otter and Briels, 1998. This is possible only in very simple cases, but not in applications to macromolecules.

We solved the problem in two steps.

·        Constraint force. In 1996 we considered systems for which the value of the constrained rc is continuously changed (without releasing the constraint). For this case we established the connection between the  mean constraint force, which is always numerically available, and the "mean force" of thermodynamic integration that determines the free energy difference: they are identical up to the sign (-1). This finding allows the easy evaluation of simulations. It turns out that the very problem can be slow relaxation and slow convergence of the mean force.

 

·        Correction. It was clear that a correction had to be found in order to adapt the free energy profile to the case without constraint. The constraint should be used merely to enable computation. Corrections had been proposed that not only were complicated, but also again entailed the old problem of partial derivatives with respect to a new complete set. In 2003 we could prove that the correction can be given a concise shape lacking of those problems:

A = free energy, r = reaction coordinate, l = constraint force, z = Fixman determinant

 

The correction is the determinant of the well-known Fixman matrix. For interesting classes of rc’s it can be shown to vanish. When other coordinates are constrained in addition to the rc the multiple constraints have to be regarded in the matrix. Even so it is a constant in many cases of practicable interest and can be omitted.

 


References:

 

Schlitter, J. and Klähn, J. The free energy of a reaction coordinate at multiple constraints: A concise formulation. J. Mol. Phys. 101, 2003, 3439-44, 2003

 

Schlitter, J. and Klähn, M. A new concise expression for the free energy of a reaction coordinate J. Chem. Phys. 118 (5), 2003, 2057-60, 2003

Schlitter, J., Swegat, W. and Mülders, T. Distance- type reaction coordinates for modeling activated processes. J. Mol. Mod.  7: 171-177, 2001

Mülders, T., Krüger, P., Swegat, W., and Schlitter, J. Free energy as the potential of mean constraint force. J. Chem. Phys. 104:4869-70, 1996.

 


  <<<< back
    2004-03-25

 

Layout (c) Marc Saric 1998