Chinese Journal of Magnetic Resonance ›› 2015, Vol. 32 ›› Issue (1): 67-77.doi: 10.11938/cjmr20150108

Previous Articles     Next Articles

L1-Norm Support Vector Machine and Its Application in Metabonomics

DING Guo-hui1,2,4,SUN Jian-qiang1,2,4,WU Jun-fang1,2,3,HUANG Shen1,2,4,DING Yi-ming1,2*   

  1. 1. Wuhan Institute of Physics and Mathematics, Chinese Academy of Sciences, Wuhan 430071, China;
    2. Key Laboratory of Magnetic Resonance in Biological Systems, Wuhan Center for Magnetic Resonance(Wuhan Institute of Physics and Mathematics, Chinese Academy of Sciences), Wuhan 430071, China;
    3. State Key Laboratory of Magnetic Resonance and Atomic and Molecular Physics (Wuhan Institute of Physics and Mathematics, Chinese Academy of Sciences), Wuhan 430071, China;
    4. University of Chinese Academy of Sciences, Beijing 100049, China
  • Received:2014-06-20 Revised:2015-01-12 Online:2015-03-05 Published:2015-03-05
  • About author:*Corresponding author:DING Yi-ming, Tei: +86-27-87199080, E-mail: ding@wipm.ac.cn.
  • Supported by:

    国家青年自然科学基金资助项目(21105115).

Abstract:

Metabonomics analyzes metabolite profiles in living systems and its dynamic responses to changes of endogenous (i.e., physiology and development) and exogenous(i.e., environment and xenobiotics) factors. Pattern recognition plays an important role in data-processing in metabonomic. L1-norm support vector machine (L1-norm SVM) is an accurate and robust method in pattern recognition, but not widely used in metabonomics. In this study, we used L1-norm SVM to analyze metabonomic data obtained from mice infected by schistosomiasis. It was shown that L1-norm SVM had better performance than
orthogonal partial least squares (O-PLS) in terms of clustering and feature selection. The results also showed that support vector machines have great potential and prospects for data-processing in metabonomics.

Key words: pattern recognition, L1-norm support vector machine, orthogonal partial least squares, metabonomics, nuclear magnetic resonance

CLC Number: