Multivariate Analysis Application Metabonomics - Sponsored Whitepaper

Metabonomics is an extension of genomics and proteomics which deals with quantitative measurement of metabolomes, or molecular profiles, and their response to external factors. Metabonomics can lead to more efficient drug discovery, specialized treatments, and targeted diets. Multivariate data analysis is essential in metabonomic studies. The data presented here originates from 1H NMR analyses of urine from thirty-two rats fed a diet containing one of the two derived onion by-product fractions: an ethanol extract and the residue. This study is based on the following publication: An exploratory NMR nutri-metabonomic investigation reveals dimethyl sulfone as a dietary biomarker for onion intake,[1]. A 24-hour urine sample was analyzed using 1H NMR spectroscopy in order to investigate the effects of onion intake on the rat metabolism.

Application of variable selection by the uncertainty test in PLS regression proved to be able to identify two dietary biomarkers for onion intake. These were identified as dimethyl sulfone and 3- hydroxyphenylacetic acid.

Being able to detect specific dietary biomarkers is highly beneficial in the control of nutritionally enhanced functional foods.

This study reveals two biomarkers for onion intake. They were found by PLS regression on different chemical areas of the NMR spectrum followed by variable selection based on an uncertainty test. Those markers were confirmed to be able to predict the onion content in the rats' diet using a test set of 4 samples.

This example of metabonomics study shows how multivariate analysis can be used to find signals of interest in a larger signal, and that one can develop a predictive model for the onion content in the diet from the NMR spectrum of urine.