Protein Folding Implementation of the Simulated Annealing Algorithm on Simple Three-Dimensional Models

Abstract

How an arbitrary coil of amino acids folds into its functional structure is known as the protein folding problem. Since the underlying mechanisms that guide protein folding in nature are widely unknown, simpli ed models are studied. Many of these models have energy levels as the focal point in order to nd the native state and may ignore other relevant constraints. While these simpli ed models may seem too trivial to have any resemblance to the physical reality, they can be used to explore concepts and ideas that may lead to further insights on how proteins fold. This thesis studies the use of simulated annealing optimization techniques to find low energy states in simple lattice and off-lattice models. A certain emphasis is placed upon looking for patterns in the results emerging. One simple off -lattice model and two lattice models are considered, a cubic lattice and a face-centered cubic lattice. Compared to the optimal energy, low energy conformations of 48- residue chains are found in reasonable time. It is concluded that while the method can not be said to exhibit the behavior of nding one consistent native state each time it is run, patterns do emerge in the results.