Polarizable force fields for molecular dynamics simulations of biomolecules

Molecular dynamics simulations are well established for the study of biomolecular systems. Within these simulations, energy functions known as force fields are used to determine the forces acting on atoms and molecules. While these force fields have been very successful, they contain a number of approximations, included to overcome limitations in computing power. One of the most important of these approximations is the omission of polarizability, the process by which the charge distribution in a molecule changes in response to its environment. Since polarizability is known to be important in many biochemical situations, and since advances in computer hardware have reduced the need for approximations within force fields, there is major interest in the use of force fields that include an explicit representation of polarizability. As such, a number of polarizable force fields have been under development: these have been largely experimental, and their use restricted to specialized researchers. This situation is now changing. Parameters for fully optimized polarizable force fields are being published, and associated code incorporated into standard simulation software. Simulations on the hundred‐nanosecond timescale are being reported, and are now within reach of all simulation scientists. In this overview, I examine the polarizable force fields available for the simulation of biomolecules, the systems to which they have been applied, and the benefits and challenges that polarizability can bring. In considering future directions for development of polarizable force fields, I examine lessons learnt from non‐polarizable force fields, and highlight issues that remain to be addressed. WIREs Comput Mol Sci 2015, 5:241–254. doi: 10.1002/wcms.1215 This article is categorized under: Structure and Mechanism > Computational Biochemistry and Biophysics Molecular and Statistical Mechanics > Molecular Mechanics Molecular and Statistical Mechanics > Molecular Dynamics and Monte-Carlo Methods