Gene differential expression analysis for leukemia based on relative risk

The occurrence and development of tumor is usually caused by gene mutation and abnormal expression, thus, differentially expressed genes associated with tumor provide a significant reference in the process of gene therapy of tumor. In this paper, we propose a gene differential expression analysis method based on relative risk to extract differentially expressed genes. The proposed method was tested in leukemia gene expression data set. The experimental results show that the method can extract significantly differential expression genes related with leukemia, and improve the classification accuracy of three state-of-the-arts of classifiers: C4.5, Naive Bayes and SVM. Furthermore, compared with SAM (Significance Analysis of Microarrays) method, the proposed method is more accurate for classification.