The aim of untargeted metabolomics is to comprehensively profile the full range of metabolites present within a biological sample. Several experimental factors strongly influence success: extraction method, separation strategy, ionization mechanism, mass spectrometer, and informatic software. Which combination of technologies is best? Although counting signals (i.e., metabolomic features) has commonly been used as a standard metric of comparison, this approach can be misleading and provide inaccurate results due to the significant number of artifacts in metabolomic data sets.
Here we present an alternative strategy for benchmarking metabolomic technologies called credentialing. The credentialing approach facilitates removal of artifactual features without the resource-intensive burden of structural identification. Some surprising insights gained from the application of credentialing will be presented.