Special-Purpose Hardware for Computational Biology
Computer simulations of proteins and other biomolecules have been the subject of significant research over the past few decades for a variety of reasons. These include improved understanding of the structure–function relationships within a protein; engineering of biomolecules to meet specific targets such as improved stability or binding specificity; and gaining a deeper insight into the interactions between proteins, and within proteins themselves, to better understand the mechanics of these biological entities.
Specifically, both molecular dynamics (MD) and Monte Carlo (MC) simulation techniques have been applied to study the conformational degrees of freedom of biomolecules. Biomolecules of interest for therapeutic purposes, such as antibodies and antibody derivatives, are generally very large in size, with 3N degrees of freedom, where N is the number of atoms in the protein. Both simulation techniques typically employ physics-based “force fields” to computationally estimate the interactions between the atoms and propagate motion in discrete intervals, which collectively describe one possible pathway of conformational change undertaken by that protein. While MD techniques rely on Newton’s laws of physics to propagate motion and are a wholly deterministic algorithm, MC techniques rely on random displacements followed by a scoring function of some sort to determine the propagation of motion. In this chapter, ...