OS-PCA: Orthogonal Smoothed Principal Component Analysis Applied to Metabolome Data.

ABSTRACT: Principal component analysis (PCA) has been widely used in metabolomics. However, it is not always possible to detect phenotype-associated principal component (PC) scores. Previously, we proposed a smoothed PCA for samples acquired with a time course or rank order, but hypothesis testing to select significant metabolite candidates was not possible. Here, we modified the smoothed PCA as an orthogonal smoothed PCA (OS-PCA) so that statistical hypothesis testing in OS-PC loadings could be performed with the same PC projections provided by the smoothed PCA. Statistical hypothesis testing is especially useful in metabolomics because biological interpretations are made based on statistically significant metabolites. We applied the OS-PCA method to two real metabolome datasets, one for metabolic

SUBMITTER: Yamamoto H 

PROVIDER: S-EPMC7999099 | biostudies-literature | 2021 Mar

REPOSITORIES: biostudies-literature