The data substrate of exposome intelligence: an interoperability profile for untargeted metabolomics

Abstract

Untargeted metabolomics, anchored in high-resolution mass spectrometry, has matured into the central analytical platform of human exposomics. It can capture endogenous biology, diet, drugs, microbial chemistry, environmental contaminants, and their transformation products from a single biological sample. Yet exposome science remains stubbornly single-study: most untargeted exposomics publications stand alone, featuring tables, partial annotations, and semi-quantitative intensities that cannot be combined across cohorts. The bottleneck is no longer instrumentation or annotation; it is interoperability. Existing standards, including MSI, mQACC, BP4NTA, NORMAN, MERIT, mzML, mzTab-M, ISA-Tab, the Universal Spectrum Identifier, RefMet, ChEBI, MetaboLights, Metabolomics Workbench, GNPS/MassIVE, and the emerging GA4GH human exposome data standards, cover the necessary ingredients but do not yet compose a single, executable profile. I argue that the next stage of exposomics must move from FAIR deposition to meta-analysis-ready evidence: a four-layer stack of acquisition comparability, machine-readable reporting, evidence-aware annotation, and standardized summary statistics, validated by a living community benchmark. Cumulative exposome science depends on it.