Mechanistic Interpretation of Toxicology Metabolomics Data

Abstract

The toxicological interpretation of metabolomics data remains challenging, mainly due to the lack of relational knowledge of metabolic pathway perturbations and adverse outcomes. Here we propose an approach focused on the associative events defined by the adverse outcome pathway (AOP) concept to derive adverse effect predictions from toxicology metabolomics data sets by combining knowledge-driven hypothesis generation and data-driven hypothesis testing. By assessing the associative key events in an AOP, a list of plausible metabolite perturbations can be created, aiding the interpretation of the list of observed metabolite perturbations or differentially abundant metabolites (DAMs). We describe the critical steps of the interpretation and certainty assessment of the effect prediction using protoporphyrinogen oxidase (PPO) inhibition as an example. The approach could serve as a stepping stone toward creating a database of validated, toxicologically meaningful associative event signatures that can be deployed both in (early stage) research of chemical product development and in regulatory chemical safety assessment for hazard identification.