Research On Multi-objective Particle Swarm Optimization Algorithm For DNA Encoding

DNA computing is a parallel computing model based on DNA molecules,which is a novel and potential interdiscipline developed on the basis of computational science and molecular biology.Without any external conditions,the designed DNA molecules can accurately self-assemble and solve complex combinatorial optimization NP complete problems.However,if the quality of DNA molecules is not high,the non-specific hybridization or inconsonant melting temperature between DNA molecules are likely to lead to the failure of the DNA computing process.In order to improve the efficiency,reliablility and the scale of the problem that can be solved,the high-quality DNA encoding is required.The design of DNA molecules needs to meet various Hamming distance constraints,thermodynamic constraints and biological constraints,and is a typical multi-objective optimization problem.However,the fitness function values of the traditional multiobjective optimization problem candidates are usually independent of each other,namely,the fitness function value of any candidate solution is only related to the objective function and has nothing to do with other candidate solutions.However,the answer to DNA encoding problem is a set of mutually constrained DNA molecules that do not exist nonspecific hybridization between them,and their fitness function values are calculated depending on other DNA molecules.Therefore,the traditional multi-objective optimization algorithm can not be well applied to solve DNA encoding problems.In this paper,a novel dynamic multi-objective Particle Swarm Optimization algorithm for DNA encoding is proposed,which is based on the characteristics of multiobjective particle swarm optimization and DNA encoding.The algorithm maintains an optimal population and an elite population.At each iteration,only the particles in the optimal population are updated.The elite population of the next generation is selected from the optimal population and the elite population by the dynamic elitist selection algorithm based on the minimum Manhattan distance.Until the maximum number of iterations is reached,the particles in the elite population are a set of DNA encoding sequences generated by the algorithm.Seven DNA sequences with length of 20,fourteen DNA sequences with length of 20 and twenty DNA sequences with length of 15 are generated by the proposed algorithm and compared with the results of the known literature about DNA encoding algorithms.The experimental results show that the algorithm proposed in this paper is reliable and efficient in DNA encoding,which can generate high quality DNA molecules and effectively improve the scale and reliability of DNA computing.