Tumor Identification Based on Gene Expression Profiles and the Search About Extraction of the Feature Genes
Lixin Gao · Shuxue de shijian yu renshi · 2011
Based on the problem of extracting information gene expression profiles,we use the wilcoxon rank method to removeunrelated genes.Based on the characteristics of high and low levels of gene expression,the bilinear regression model is established on the high/low expression levels.The 19 feature genes is extracted based on the Residual analysis. The heuristic breadth-first search algorithm is used to search for the optimal gene subset, and then to determine the gene tag of colon cancer.There is good effect by the inspection for classification results using support vector machine.