Molecular Simulation and Systems Biology - Systems Biology: Networks, Models, and Applications - page 74

such as electron polarization and charge transfer. A number of force
fields have been developed for important and common classes
of biomolecules. These have been hand-optimized to give more or
less realistic approximations to the intermolecular interactions for a
number of important purposes. This effort has engaged hundreds
of researchers over a period of close to twenty years. And still, when
using a force field or any other approximation to the intermolecular
interactions, one must always be careful to verify that the interaction
model is adequate for the purpose at hand.
Key Challenges in Molecular Simulation
In spite of the large number of approximations involved, and the
availability of very powerful computers, building, performing, and
interpreting simulations is not a trivial task. There are several critical
challenges that must be met in the field of molecular simulation.
Both MD and MC techniques require a model that specifies the
interactions between particles in the system. One of the most signifi-
cant challenges for the field has to do with the development of accurate
representations of the molecular interactions. This is sometimes
referred to as force field quality, but it appears also in the context of
the various quantum mechanical approaches where it could relate to
the degree of accuracy associated with the choice of quantum method
and basis set. Quantum mechanical approaches can be limited in terms
of the quality of the functional (in the case of density functional theo-
retical approaches), the size and quality of the basis set, treatment of
solvation effects, temperature, and boundary conditions.
For classical modeling, force field approximations to the molecular
interactions come in many flavors. Some of these include treatment of
electronic polarization effects and some do not. The ones that do not
include polarization, sometimes known as fixed charge force fields,
have been developed and tuned over many years for various materials.
For protein simulations, there are a handful of fixed charge force fields,
including CHARMm [9], OPLS-AA [10], AMBER [11], and GROMOS [12].
These have reached a certain degree of maturity, and many experts
in the field know which force fields are adequate for the study of vari-
ous types of biological processes. However, some very important
phenomena are inherently quantum mechanical in nature and the use
of force fields is inappropriate.
Related to force field quality is the method used to represent the
solvent in simulations. There are some approaches that involve treat-
ing the electrostatic effect of the solvent on a solvated molecule as if
the solvent were a continuum dielectric. These are known as implicit
solvent models. Usually used in conjunction with continuum solvent
models is Langevin dynamics, where the dynamical effects of the
solvent are emulated by imposing short random impulses on the