Multivariate analysis of 1H NMR spectra for saponins from Quillaja saponaria Molina

The bark of the South American tree Quillaja saponaria Molina contains at least 100 different saponin structures. A set of known structures were classified by partial least squares regression-discriminant analysis (PLS-DA) on spectra from 1H NMR. PLS-DA was found to detect structural variations in the different parts of the saponins, encoded in the spectra. The models obtained were used to classify a test set of known structures. The amount of saponin material used for the NMR spectroscopy was only 200µg/sample. This means that the outlined method is an interesting tool in the structural analysis of minor saponin components from Quillaja. This non-destructive method could also be applied on and facilitate the analysis of other natural products and metabolites since only minor amounts of sample are needed.