混合法用于基因芯片数据的分类分析

A Combined Method Applied to the Classification of Microarray Gene Expression Data

目的研究广义似然比检验(GLRT)与支持向量机(SVM)相结合的混合法在基因芯片表达数据分类研究中的应用效果。方法结合Golub研究的白血病基因表达数据,利用GLRT寻找白血病表达水平存在显著性差异的基因,再使用SVM对白血病进行分类。结果基于GLRT鉴别得到的特征基因有很强的代表性,SVM分类的效果明显优于其他方法。结论混合法在解决数据量大、维数高、样本量小、非线性问题中具有很强的优势,可以有效的用于基因芯片表达数据的分类研究。

Objective Discuss the effect of the generalized likelihood ratio test （GLRT） and support vector machine （SVM） on the classification of gene expression data. Methods According to the expression profile data of Golub et al, take on a GLRT approach to find different expression gene at the expression level, and take on SVM to classify the leukemia. Results Feature gene selection based on the GLRT approach is more representatively, and the efficiency of SVM is superior to other methods. Conclusion There are lots of potential advantages of the combined method in solving the problem of classification in the nonlinear, high dimension and limited-case samples, and the combined method could be effectivelv used in the field of classification of gene expression data.