Chemistry Reference
In-Depth Information
barriers found in the surface of potential energy. The picture of the dynamics
molecular conformations can be provided by this procedure.
A static picture can be provided by using a procedure based on simple energy
minimization. A complete description of molecular motions on the sub-
picosecond time scales is obtained by this procedure providing important
information for the system investigated. This information may be only partially
accessible through experimental techniques. A knowledge of the molecule's
dynamics can yield a complete understanding of the structure-function
relationship.
It is advisable to do an energy minimization before starting a MD simulation. This
allows the energy to equilibrate among the constituent atoms, remove van der
Waals interaction, avoid/reduce local structural distortion and unstable
stimulations, kinetic energies and thermal noise in the structures and potential
energies.
These minimization methods are classified as first derivative whereas only the
first derivative of the potential energy function is calculated. In the second-
derivate methods, both the first and second derivatives are calculated as done in
the Newton-Raphson method [558]. The steepest descent is a simple method of
minimization. The conjugate gradient method minimizes the function of energy
potential in successive steps in order to achieve the minimum point whereas
information is saved to guide subsequent steps. Hessian matrix techniques are
used in the Newton-Ralphson methods with a high computational cost.
When we do molecular dynamics simulations, there is a computation of the
molecular potential energy taken as a sum of energy terms. There is a description
of the deviation (away from equilibrium values) of torsion angles, bond lengths
and bond angles. Electrostatic and van der Waals interactions are also described
by other terms. It is important to carefully determine the equilibrium bond angles,
partial charge, bond lengths, Van der Waals parameters and force constants. This
is obtained by fitting to high-level quantum mechanical calculations or
experimental data. Combination of methods for free energy predictions with MD
simulations yields a good strategy for identifying interactions for drug design.
