High-throughput mass spectrometry-based metabolomics for large-scale population studies: technical barriers and recent innovations.

Large-scale metabolomic studies are needed to identify and validate biomarkers for their successful translation in emerging population health and precision medicine initiatives. Yet, untargeted mass spectrometry (MS)-based metabolomics remains hampered by low sample throughput, limited coverage, complicated data processing, poor standardization, and high operating costs. These challenges also contribute to a reproducibility crisis, with most reported biomarkers not independently replicated, given variations in study designs, sample workup protocols, instrumental methods, and data pre-processing workflows. This review will focus on new advances in high-throughput MS platforms and computational software with emerging artificial intelligence/machine learning tools that are scalable, faster, and less prone to error. Recent progress to improve data harmonization in population-level metabolomic, lipidomic, and exposomic studies (n > 1000) will also be discussed. This will include strategies to correct for long-term batch effects, improve metabolite identification, and derive biochemical insights into disease mechanisms when integrated with multi-omic studies.