Protein folding is a relevant computational problem in Bioinformatics, for which many heuristic algorithms have been proposed. This work presents a methodology for the application of differential evolution （DE） to t...Protein folding is a relevant computational problem in Bioinformatics, for which many heuristic algorithms have been proposed. This work presents a methodology for the application of differential evolution （DE） to the problem of protein folding, using the bi-dimensional hydrophobic-polar model. DE is a relatively recent evolutionary algorithm, and has been used successfully in several engineering optimization problems, usually with continuous variables. We introduce the concept of genotype-phenotype mapping in DE in order to provide a mapping between the real-valued vector and an actual folding. The methodology is detailed and several experiments with benchmarks are done. We compared the results with other similar implementations. The proposed DE has shown to be competitive, statistically consistent and very promising.展开更多
基金This work is supported by the Brazilian National Research Council under Grant No.305720/04-0.
文摘Protein folding is a relevant computational problem in Bioinformatics, for which many heuristic algorithms have been proposed. This work presents a methodology for the application of differential evolution （DE） to the problem of protein folding, using the bi-dimensional hydrophobic-polar model. DE is a relatively recent evolutionary algorithm, and has been used successfully in several engineering optimization problems, usually with continuous variables. We introduce the concept of genotype-phenotype mapping in DE in order to provide a mapping between the real-valued vector and an actual folding. The methodology is detailed and several experiments with benchmarks are done. We compared the results with other similar implementations. The proposed DE has shown to be competitive, statistically consistent and very promising.
