An Improved Method of Extracting and Classifying DLBCL Information Genes

Abstract

The extraction of tumor information genes and the processing of gene expression profile data is a very important step in the study of gene expression profile, which is of great significance to the diagnosis of patients. In this paper, a novel method for Diffuse Large B-cell Lymphoma (DLBCL)information gene extraction and classification is proposed based on graph theory. Firstly, the expression of each gene under different conditions is mapped to make it easy to use the knowledge of graph theory to mine rules. Then singular value decomposition (SVD) was used to obtain the spectral information of the map and characterize the gene expression rule. The information gene subset was selected according to the calculation of cosine Angle and distance between the map and the ideal template. Finally, experiments are carried out on two public DLBCL data sets, and the experimental results show that the classification accuracy is above 85% no matter how many information genes are selected or the parameters of the classifier are adjusted. The optimal classification accuracy is 98.7%, which is satisfactory. The expression patterns of information genes related to DLBCL type recognition were presented to assist tumor specialists in identifying and treating DLBCL.