Metabolites, signifying a complex interplay between genotype, behaviour, and environment, provide a unique readout of human health and disease. This website is a public repository of human metabolome-phenome atlas, providing over 1.4 million associations between 313 plasma metabolites and a comprehensive range of human phenotypes (comprising 1,386 diseases and 3,142 traits) across 274,241 longitudinally followed (median 14.9 years) participants from UK Biobank. The present atlas was approved by UK Biobank under application number 202239 and 19542.

The atlas provides epidemiological associations between metabolites and deep human phenotypes of diseases, health-related traits and imaging traits. The atlas also maps distinct metabolite variational patterns to disease development, revealing key time points of metabolic changes over a decade preceding onset. Machine-learning-based Metabolic Risk Score signatures the top metabolites as potential biomarkers for prevalent and incident diseases discriminations. In search of promising therapeutic targets, the atlas uncovers potentially causal-related metabolite-disease pairs and shared genetic determinants.

Detailed information regarding the data source and analytical methods utilized can be located in the following publication. Kindly cite the paper should you choose to employ any findings from this website. Please note that the paper is currently under review and will be updated accordingly upon its publication.