Automated Annotation of Microbial and Human Flavonoid-Derived Metabolites

Flavonoids are a class of natural compounds essentially produced by plants that are part of animal and human diets and have assumed health-promoting benefits. Upon human consumption, these flavonoids are to a modest extent absorbed in the small intestines. The major part arrives in the colon where the microflora utilises and converts the flavonoids to a wide range of products. Many of these products are absorbed in the major intestines and subsequently metabolised by the host. To understand the impact of the microflora on the metabolism and possible effects on human health, complete (and quantitative) identification of the microbial as well as human metabolic conversion products of flavonoids is required. This is a challenging task, as these bioconversion products are often present in relatively small amounts, making classical identification strategies based on (accurate) mass information or nuclear magnetic resonance, not straightforward. In the absence of reference compounds, annotation of a component may be achieved by detailed expert evaluation, e.g. by searching for similar fragmentation patterns in spectral databases of known compounds. However, such manual analysis is a tedious task, and in advanced metabolite profiling experiments, with large numbers of unknown metabolites, this is a major bottleneck. Therefore, new strategies are needed for quick and reliable identification of the diverse range of molecules in complex matrices (faeces, blood, urine). Intelligent software for annotation and identification of unknowns is crucial to fully exploit complex datasets. We developed a new software tool (MAGMA) for (sub)structure-based annotation of LC-MSn datasets which, combined with a newly established database for phenolic molecules (MetIDB), enables semiautomated identification of flavonoid derivatives.