INITIAL EVENTS OF PROTEIN FOLDING FROM AN INFORMATION-PROCESSING VIEWPOINT

The initial events of protein folding can be described by the time evolution of a neural network through the use of polymer-kinetics theory, where the fundamental variables are residue contacts. The contact energies are analogous to chemical activation potentials. The resulting recurrent neural network behaves like a partially frustrated spin system interacting with a random external field. A dynamical phase diagram is derived from numerical studies of the model, and the basins of attraction for the network are studied.