Interpretation of multivariate metabolomic models through network-guided perturbation-based explanations.

Introduction: Multivariate modeling is crucial for uncovering complex patterns in metabolomic data, yet the interpretability of such models remains a major challenge.

Methods: Here, we propose a network-guided framework that enhances perturbation-based explanations by grouping metabolites according to communities identified in metabolic networks, rather than relying on predefined pathways. The approach is applied to postprandial plasma metabolomic data as a model example and using a metabolic network including KEGG metabolites and enzyme-catalyzed reactions in which they participate.

Results and conclusion: Results show that the use of metabolite communities derived from network representation in perturbation-based analysis of multivariate models, serves as a complementary tool for their biochemical interpretation, that might extend it beyond fixed, established pathways. The strategy is model-agnostic and readily transferable across omics domains and multivariate methods, offering a new tool for model interpretability and hypothesis generation in complex biological datasets.