Untargeted Metabolomics in Fermented Food Systems

Fermented foods are chemically complex systems in which substrate composition, microbial community dynamics, and physicochemical conditions interact to generate thousands of metabolites across diverse chemical classes. Conventional targeted analytical approaches quantify predefined compounds with high precision but operate within a restricted chemical space, systematically excluding emergent features central to product identity, safety, and sensory character. Untargeted metabolomics addresses this limitation by capturing global chemical fingerprints of fermented matrices, enabling discovery-driven investigation across a broad fraction of the metabolome. This review examines the application of untargeted metabolomics across key research areas in fermented food science, including fermentation monitoring, microbial interactions, flavour development, process optimisation, post-fermentation stability, and safety assessment. Across these domains, untargeted approaches reveal system-level metabolic relationships beyond the reach of targeted analyses, while also presenting interpretive challenges. A central limitation is the annotation bottleneck: despite high feature detection rates, only a small fraction of signals are structurally identified, constraining mechanistic interpretation and cross-study comparability. Additional challenges in data processing, statistical validation, and interlaboratory reproducibility further limit data interpretation. Addressing these constraints through improved spectral libraries, standardised workflows, and integration with complementary omics is essential for advancing untargeted metabolomics towards robust knowledge generation in fermented food systems.