The results of multivariate analyses of sets of spectra (e.g. by PCA or PLS) are scores plots, which represent the relationships of samples, and loadings plots, which represent relationships of signals. Typically the assignment of significantly changed signals to significantly changed metabolites is a tedious manual task. Recently, a new technique has been developed for an easy automated assignment of signals to metabolites (Dieterle, Ross, Schlotterbeck and Senn, Metabolite Projection Analysis). This new technique is a direct extension of common multivariate analyses. The method bases on the projection of spectra of metabolites into multivariate models.
This method allows a quick and easy assignment of significantly changed signals to signicantly changed metabolites, whereby only metabolites can be assigned, for which spectra are available. In order to check for false positive assignments and for new unknown metabolites, further investigations are always necessary.
Assignment of significantly changed signals to significantly changed metabolites by the metabolites projection analysis, which is a direct extension of multivariate models.