| Protein folding problem is one of the core research areas in bioinformatics, and describes how an amino acid sequence folds into a specific spatial protein. Due to the critical positions of its studies in the process of transmitting bio-information and the living organisms discovered, the intelligent computation has been playing more and more important role in modeling, prediction and optimization for protein folding systems. In this thesis, a novel optimization algorithm, so-called Hysteretic Optimization (HO) is applied to dealing with 3D protein folding problem with lattice model. An effective improved method to cope with the 3D protein folding problem is proposed, according to the characteristics of HO and 3D-protein folding problem.This study involves the first application to solving three-dimensional protein folding problem with HO, which is inspired by ac demagnetization (ACD) procedure in magnetic systems proposed in 2002 by Zarand and his co-workers. In this study, the solutions start from benchmark data inputs, problem formulation, to the algorithm development and implementations. We conclude HO consists of a number of building blocks, and the vital factor of employing HO to protein folding problem is to make proper definitions on the key ingredients of HO in order to establish the relationships connecting with the HO principle, protein folding structure, and the energy of permutation as well.Then based on the numerous previous publications, a proposed modified HO algorithm is developed and successfully implemented for studing 3D protein folding problems. And the benchmark based numerous simulation results show the efficiency of the proposed HO method. Especially, for the protein formed by 64 amino acids, we first find a conformation with energy-57 smaller than the minimal value -56 before. Finally, the applications of hysteresis optimization algorithm in 3D protein folding problem is summarized, and the future research work is proposed in concluding remarks.. |
PDF Full Text Request