Basin Hopping as a General and Versatile Optimization Framework for the Characterization of Biological Macromolecules
Figure 1
(a) A short polypeptide chain of amino acids, alanine, lysine, lysine, and valine, is shown. The backbone atoms shared by all amino acids are N, , C, and O. Side-chain atoms unique to types of amino acids are in gray. The backbone () dihedral angles are annotated over the chain. (b) A model energy surface is illustrated, adapted from [25]. The surface is funnel-like but rugged. The native state at the bottom is denoted by N. Conformations associated with it (obtained from experiment) are illustrated for a particular protein molecule. (c) The BH framework essentially converts the function into a stepwise one. The perturbation and local optimization components are illustrated here with differently colored arrows. A minimum is shown here which fails the Metropolis criterion and is thus not accepted, prompting a new perturbation move.