Abstract: In this study, a technique that combined both analysis of variance(ANOVA) and partial least squares-discriminant analysis(PLS-DA) was used to compare the urine 1H NMR spectra of healthy people from a vegetarian and omnivorous population. In ANOVA/PLS-DA, the variation in data was first decomposed into different variance components that each contains a single source of variation. Each of the resulting variance components was then analyzed using PLS-DA. The experimental results showed that ANOVA/PLS-DA is efficient in disentangling the effect of diet and gender on the metabolic profile, and the method could be used to extract biologically relevant information for result interpretation.

Key words: Multi-factor analysis, Analysis of variance(ANOVA), Partial least squares-discriminant analysis(PLS-DA), NMR-based metabolomic