数据挖掘技术在生物信息学基因变异规律研究中的应用

目的聚类分析中适宜类别数的确定和结果的验证一直是难以解决的问题，而在生物信息学研究中这一矛盾尤为突出，本文试图将数据挖掘技术引入这一领域，以期能有所帮助。方法以甲型流感病毒的H3序列为例，按照数据挖掘的思想将其拆分为训练集和验证集，然后使用两阶段聚类法和自组织图进行聚类分析，利用验证集对聚类结果进行验证，并进行各类的特征描述。结果两阶段聚类法可自动搜索适宜的类别数，两种聚类方法结果相互验证，以及验证集的结果验证都证实了聚类结果的准确性。结论数据挖掘方法体系中的智能聚类技术可以满足基因序列数据聚类问题的需求，其相关技术可较好的解决类别数判定、结果验证等问题，值得在该领域中推广。

Application of Data Mining in Gene Mutation Research of Bioinformatics

Objective The problem of cluster number determinant and result validation,which are more server in data analysis of bioinformatics,are difficult tasks in cluster analysis.This paper is aimed to use data mining technique to handle the related issues.Methods H3 gene sequence data of Influenza A virus are invoked as example.Followed on data mining flow,data are split into train set and test set,two step cluster method and self organization map are applied for model training,then test set are used to verify model's robust,cluster feature are explored at the same time.Results Two step cluster method can find the best suitable cluster number automatically,while both analysis result of test set and cross-validation of two methods show high stabilization of the cluster result.Conclusion Data mining technical can satisfy needs for cluster analysis of gene data in bioinformatics,the related technique can resolve problem lie in it.Therefore,more widely application of data mining should be advocated in this field.