Based On Gene Expression Regulatory Networks Constructed To Improve The Association Rules And Genetic Algorithm Method

The research on gene regulatory network is one of the tasks of the post-genome informatics, which applies bioinformatics methods and techniques such as data acquisition, analysis, modeling, simulation and speculation to study complex biological networks. So it can reveal the mechanism of life in genomics level and is the forefront of life science at present as well. The purpose of this study is to establish the gene transcriptional regulation network through which we can simulate, analyze and study all the genes'expression relationships in a certain species or tissue to understand life in the framework of the system, especially the laws of information flow.Based on the construction of gene expression regulation networks in this paper, association rules algorithm was firstly improved such as adding the extra conditions for reducing the numbers of items, adopting the compression of transactions to reduce the numbers of transactions, and taking storage technology to enhance scanning speed. The initial population generating, coding/decoding approaches, fitness function and genetic manipulation in genetic algorithm are designed. The optimum selection operator was adopted to improve the searching efficiency of genetic algorithm. Then, the improved association rules algorithm and genetic algorithm were combined to establish a novel approach——ARGA (Association Rules and Genetic Algorithm) algorithm for constructing gene expression regulatory networks. Finally, the simulation experiments for Yeast gene expression data showed that the gene expression regulatory networks with important biological meanings can be constructed according to the discoverable association rules. This method inherits their respective advantages of association rules and genetic algorithm, and can search for the rich expression patterns and optimal association rules, and can avoid the deficiency of cluster that a gene can only be assigned to a certain class. The feasibility, effectiveness and practical application values for the novel approach were verified.