PROJECT SUMMARY / ABSTRACT Dysregulated metabolism underlies many of the leading causes of mortality and morbidity in the US including cardiometabolic diseases. Metabolomics studies can identify novel disease biomarkers, novel therapeutic targets, and biological pathways with pathological relevance. Emerging technologies in metabolomics allow the interrogation of large numbers of metabolites from diverse pathways. However, these approaches remain expensive and time-consuming. Applying metabolomics to very large cohorts of individuals to conduct epidemiological studies is not feasible, due to the practical challenges and costs of implementing these assays at scale. These challenges have limited discovery of novel biomarker-disease associations. We propose to address these limitations with a genetics-based “virtual” metabolite study design that will allow us to define genetic predictors of metabolite concentrations in a small population in whom the metabolite was measured, and then use these genetic predictors to impute metabolite concentrations in a large population in whom the metabolite was not measured. This approach vastly amplifies the sample size for discovery, and can rapidly identify novel biomarkers for downstream validation. The primary aims of this proposal are to: 1) construct single nucleotide polymorphism (SNP)-based predictors of circulating metabolites, and identify associations with cardiometabolic phenotypes, including type 2 diabetes and coronary artery disease; 2) validate the associations with direct metabolite measurements; 3) identify pleiotropic associations between metabolite genetic predictors and the clinical phenome. These analyses are enabled by genetic approaches that allow us to integrate data from large scale genome-wide association studies (GWAS) of cardiometabolic diseases and a collection of electronic health record linked-DNA biobanks comprising over 700,000 subjects. Innovative features of this approach include the efficiency and scale of the analysis, inclusion of under-represented and vulnerable populations and implementation of a re-usable and scalable analytical framework that will accelerate biomarker discovery and implementation. Upon completion of this project, we will construct a publicly accessible online resource of metabolite-disease associations that will be available to researchers as a source for both hypothesis testing and generation. Ultimately, these studies will advance the field of metabolomics by rapidly advancing the process of linking metabolites to clinically-relevant diseases.