Abstract: In view of the problem of gene sequence classification, combined with fuzzy clustering analysis method that has the advantages of simple classification and high accuracy of classification result. Based on the original Markov chain model of the gene clustering method, and the feature nucleic acid base pair interactions has been taken into account to calculate the distances. A characteristic distance matrix with dual properties is obtained. The method gets fuzzy similar matrix and dynamic clustering graph by suing fuzzy clustering, which achieves classification of the DNA sequence. Selecting 16 species p53 gene sequences including humans to study the relationship between species closer, the more easily become a class. The results are consistent with those previous analyses, which illustrates the utility of the approach. In addition, the clustering results of clustering with dual property method are compared with the clustering results of single property method and find that the dual property method is more accurate.

Key words: fuzzy clustering, Markov chain, interaction, dual property

摘要： 针对基因序列分类的特点，结合模糊聚类分析方法，在原来的Markov链模型基因聚类方法的基础上，引入核酸碱基对的相互作用，得到具有双重性质特征的距离矩阵，并根据模糊聚类分析方法得到模糊相似性矩阵和其动态聚类图，从而实现基因序列的分类。通过对包括人类16个物种的16条p53基因序列进行模糊聚类得出，物种关系越相近，更容易聚成一类。此外，还检验双重性质的矩阵方法与原来的单一性质方法作聚类结果对比，发现具有双重性质的方法更准确。

关键词: 模糊聚类, Markov链, 相互作用, 双重性质