Monte Carlo lattice dynamics and the prediction of protein folds

Abstract

The general solution to the protein folding problem demands that two very difficult problems be concomitantly solved [1]. An energy function whose global minimum is in the native conformation of the protein must be developed [2]. Simultaneously, an efficient strategy to search through the myriad of local energy minima for the desired global minimum must be formulated [3]. One way to attack both problems is to reduce the complexity of the model being considered [4]. Rather than treat the model at the level of atomic detail, the representation of the protein can be simplified. Various extents of this simplification have been explored. They range from highly simplified models that treat the native conformation of proteins as points on a small cube to high coordination lattice models that describe the native conformation of proteins with high geometric fidelity [5–14]. While highly idealized models have been useful in providing a number of qualitative insights into some general features of protein folding [10,15,16], they cannot be used to fold a real protein. This chapter focuses on results from high coordination lattice models of proteins that have been developed over the past several years and which are complementary to the simplified model studies [17–31]. These high coordination models not only provide insights into the thermodynamics of the protein folding process [25], but in a number of cases can predict the native conformation of a number of proteins at the level of 2–4 Å root-mean-square deviation (rms) from native [20,21].