Auto-selection of Informative Gene for Multi-class Tumor Gene Expression Profiles
多类别肿瘤基因表达谱的自动特征选择方法

In microarray analysis, the selection of informative gene is an essential issue for tissue classification and successful treatment because of its ability to improve the accuracy and decrease computational complexity. The ability of successfully distinguishing tumor from normal tissues using gene expression data is an important aspect of this novel approach for cancer classification. In this paper, a non-parameter method for autonomous selection of informative gene was proposed for processing multi-class tumor gene expression profile,which contained 218 tumor samples spanning 14common tumor types, as well as 90 normal tissue samples, to find a small subset of genes for distinguishing tumor from normal tissues. At First, the randomness of a decision sequence was defined to measure gene importance based on the non-parameter method and filter algorithm. I}hcn correlation information entropy was used to eliminate redundant genes and selected informative feature genes. As a result, 30 informative genes are selected as markers for making distinctions between different tumor tissues and their normal counterparts. Simulation experiment results show that the selected genes arc very efficient for distinguishing tumor from normal tissues. In the end, several methods for informative gene selection were also analyzed and compared to validate the feasibility and efficiency of the proposed method for dealing with tumor gene expression profiles.