Force field models, empirical water simulation

Recently, we studied the water dependence in mixtures of water and the protonated form of Nafion [53] using both standard force field models and an empirical valence bond model to account for the Grotthuss structural diffusion mechanism of aqueous proton transport. Results showed a transition of an irregularly shaped filamentous (cylindrical) structure in the case of low water (A = 5, where A is defined as the ratio of water molecules to sulfonate groups) to a structure more in accord with the above-discussed models of nano-separation, where larger clusters form which are connected by narrow bridges. A comparison of aqueous cluster sizes indicated, for a simulation time of 30 ns, no percolating clusters for A = 5, whereas at A = 10 most water molecules were located in a connected cluster (see [53]). Other structural... 

An alternative to the computationally intensive method of developing force fields from quantum mechanics has been to use empirical potentials, either transferable potentials that are meant to be used for many compounds (c.g OPLS, AMBER, TraPPE, etc.) or specialized potentials for specific compounds e.g., SPC, " TIP4P and later models for water). Such empirical potentials have been fit (frequently using simulation) to some experimental data. The results in Figure 7 for methanethiol illustrate the potential inaccuracies of using transferable potentials. There, we see that using a potential function fit to the quantum-mechanically... 

Molecular modeling and computer simulation with empirical potential energy function (force field) are now routinely carried out to help understand and predict structures and dynamics of proteins and other macromolecules of biological relevance in water and membrane environments. After over 40 years of development, popular force fields such as AMBER, CHARMM, OPLS and GROMOS have been widely employed in biomolecular simulations. These force fields are used dominantly in highly optimized molecular dynamics... 

Typically, effective potential models are parameterized empirically by constraining the model s thermophysical properties to experimental values at ambient conditions, by adjusting those parameters in a series of short simulations [81,94,95] or during a rather longer simulation [50,85]. The most convenient choices are the configurational internal energy and the pressure to fit the energy and size force-field parameters for the pairwise repulsion-attraction terms. Additionally, almost all models have been parametrized to reproduce the structure of water obtained in the 1986 Soper-Phillips neutron diffraction experiments [96]. 

As computational power advanced and became more accessible, so did the detail incorporated into the protein models. By 1977, work had moved toward united atom representations of proteins and the use of molecular mechanics to characterize the native state dynamics of BPTI in vacuo with no water included except for four crystalline waters.The empirical energy function used in this first protein MD simulation forms the basis of today s protein force fields. 

---