Research On Double Cluster Algorithm Of Gene Expression Based On Evolutionary Multiobjective

Microarray technology can simultaneously measure the expression levels of thousands of genes.Simultaneously detect the expression levels of a large number of genes and produce a large number of microarray datasets.These gene expression data reflect the genome level under different experimental conditions.The gene expression data contains a large amount of gene activity information and cell physiological status information.How to use effective calculation methods to analyze large amounts of microarray data has become a new challenge in bioinformatics.Therefore,more effective analysis algorithms need to be researched to find out some correlations between gene expressions.On this basis,we can further explore some regulatory modes,common functions,and interactions among genes.For microarray data analysis,dual clustering is a commonly used data mining method and shows its advantages in many applications.With the analysis and application of dual clustering algorithm in gene expression data,a large number of different types of BI clustering algorithms have emerged,which play an important role in the analysis of gene expression data.In gene expression data,trend consensus dual clustering is one of the most biologically significant types of dual clustering.Many current algorithms are designed to solve this type of BI clustering.However,due to the difficulty and complexity of the problem itself,how to efficiently find the trend of consistent data in gene expression data still poses great challenges.For the analysis of gene expression data,the dual clustering method has proved to be an efficient and practical method.When designing the dual clustering algorithm,it is necessary to consider simultaneously optimizing several conflicting targets,such as higher row variance and lowest mean square residual,etc.The maximum double clustering of the target can therefore be converted into a multi-objective optimization problem.In recent decades,algorithms designed based on natural phenomena,such as evolution,particle swarms,and ant colonies,have become popular methods in data mining.By using evolutionary algorithms,the global optimal solution in the microarray gene expression data can be discovered.Optimizing several conflicting objectives such as cluster size and homology,a multi-objective evolutionary optimization algorithm was proposed to discover globally optimal biclustering in the microarray data.This article through the detailed analysis of NSGAII algorithm and MOEA/D algorithm,study their respective advantages and disadvantages.In this research,an algorithm combining NSGAII algorithm and MOEA/D algorithm is proposed.The algorithm uses the two well-known NSGAII and MOEA/D algorithms to measure their evolutionary efficiency in different stages of evolution,and then distributes them differently.The computing resources make reasonable allocation of computing resources throughout the evolutionary process to improve the performance of the algorithm.This algorithm can avoid this NSGAII diversity deficiency and MOEA/D convergence asymmetry at the same time,and can also use their respective advantages.Finally,the proposed algorithm was validated on yeast gene expression datasets and human cell datasets,and compared with the NSGA2 B double clustering algorithm using NSGAII multi-objective algorithm to solve the dual clustering problem.Through a large number of experiments,good results have been obtained.The results show that the proposed algorithm can better deal with the double clustering problem in gene expression data.