FeatureMAP enables network-guided interpretation of untargeted metabolomics

FeatureMAP improves MS1-based metabolite candidate prioritization across ten public untargeted LC-MS metabolomics datasets while reducing annotation redundancy, enabling network-guided interpretation.

• Why it matters: Untargeted LC-MS produces thousands of features, but incomplete metabolite annotation and dependence on predefined pathways constrain biological interpretation.
• What they did: FeatureMAP is a feature-centric framework that combines multi-evidence metabolite annotation, network-guided refinement, metabolic module discovery, and structured large language model functional interpretation, then tests it on ten public datasets plus semi-synthetic and controlled benchmarks.
• The result: In semi-synthetic benchmarks it achieved 92% recovery of a simulated perturbation and distinguished concurrent metabolic processes, and PhenoMAP further built a phenotype-aware knowledge graph of 12,929 dysregulated metabolic modules across 326 phenotypes.

AI summary of the abstract · may contain errors · not medical advice · read the paper.