Sparse discriminant analysis for breast cancer biomarker identification and classification

• Published in: Progress in natural science Vol. 19; no. 11; pp. 1635 - 1641
• Main Authors: Shi, Yu, Dai, Daoqing, Liu, Chaochun, Yan, Hong
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 2009; Department of Electric Engineering, City University of Hong Kong, 83 Tat Chee Avenue, Kowloon, Hong Kong, China%Center for Computer Vision and Department of Mathematics, Sun Yat-Sen (Zhongshan) University, Guangzhou 510275, China%School of Electrical and Information Engineering, University of Sydney, NSW 2006, Australia China; Department of Electric Engineering, City University of Hong Kong, 83 Tat Chee Avenue, Kowloon, Hong Kong, China; Center for Computer Vision and Department of Mathematics, Sun Yat-Sen (Zhongshan) University, Guangzhou 510275, China
• Subjects: Biomarker identification; Cancer classification; Discriminant analysis; Maximum penalized likelihood; Microarray data analysis; 乳腺癌; 判别分析; 数据分析; 生物医学; 生物标志物; 自动识别; Microarray data analysis; Discriminant analysis; Maximum penalized likelihood; Cancer classification; Biomarker identification; Mieroarray data analysis
• ISSN: 1002-0071
• DOI: 10.1016/j.pnsc.2009.04.013

Biomarker identification and cancer classification are two important procedures in microarray data analysis. We propose a novel unified method to carry out both tasks. We first preselect biomarker candidates by eliminating unrelated genes through the BSS/WSS ratio filter to reduce computational cost, and then use a sparse discriminant analysis method for simultaneous biomarker identification and cancer classification. Moreover, we give a mathematical justification about automatic biomarker identification. Experimental results show that the proposed method can identify key genes that have been verified in biochemical or biomedical research and classify the breast cancer type correctly.