Comparison of popular enrichment methods for untargeted in vitro metabolomics

INTRODUCTION: Untargeted metabolomics is a popular method by which researchers measure a large portion of the metabolites present in a biological system at once. This approach usually results in complex data sets containing tens to hundreds of thousands of observations which require sophisticated data analysis workflows. To help with the functional interpretation of the data, researchers often rely on enrichment analysis. However, little advice is available on what method to use, and, to the best of our knowledge, there is no comparison of popular approaches available for in vitro data with a focus on toxicological and pharmacological testing. OBJECTIVES: In this study, we compared three popular enrichment analysis approaches—Metabolite Set Enrichment Analysis (MSEA), Mummichog and Over Representation Analysis (ORA)—with data obtained by treating Hep-G2 cells with 11 compounds with five different mechanisms of action. We compared the results and assessed the consistency of the individual methods as well as their correctness. METHODS: Hep-G2 cells were treated with subtoxic concentrations of 11 test compounds. After preparation, samples were measured on an Elute UHPLC coupled to a timsTOF Pro (both Bruker). Spectra were processed in MetaboScape (Bruker) and annotated using spectral library search. Datasets were further processed using R and enrichment analysis was performed in MetaboAnalyst. RESULTS: Overall, we observed a low to moderate similarity between different enrichment methods with the highest similarity between MSEA and Mummichog. Further, Mummichog outperformed both MSEA and ORA in terms of consistency and correctness. CONCLUSION: In our comparison, Mummichog showed the best performance for in vitro untargeted metabolomics data.