Abstract: Metabolomics has been widely applied to life science and showing a promising perspective. Conventional statistic analysis is not applicable to the large, multivariate dataset generated by high-throughput metabolomic tool, while it's of crucial importance to analyze and interpret the dataset. This article reviews the basic methods of principal components analysis(PCA) that is popular in metabolomics study, aiming at strengthening the fundamental knowledge of PCA and standardizing the methods and procedures for data analysis.

Key words: Metabolomics, Principal components analysis(PCA), Partial least squares project to latent structure(PLS), Partial least squares project to latent structure-discriminant analysis(PLS-DA), Orthogonal partial least squares project to latent structure(OPLS)