Approved Research
Integrative Genomic and Phenomic Architecture of Plasma Metabolic Signatures
Lay summary
Cardiometabolic diseases, comprising cardiovascular diseases and diabetes, stand as paramount contributors to global morbidity and mortality. These diseases find their origins in metabolic disorders, underscoring the significance of targeting metabolic pathways for risk prediction and early intervention. The metabolome is defined as a collection of metabolites, namely small molecules involved in cellular metabolism, offering a crucial insight into human physiological processes and metabolic individuality. Nuclear magnetic resonance (NMR) metabolomics stands out as a method for uniform and quantitative assessment of an extensive array of small molecules in the plasma concurrently. Its distinctive capability lies in generating high-throughout data, exhibiting advantages of cost-effectiveness and superior repeatability when compared to mass spectrometry. Recent years have witnessed genome-wide association studies delving into the plasma metabolome, revealing intricate variant-metabolite-disease associations. These findings significantly contribute to advancing disease risk prediction and uncovering potential therapeutic targets. Contemporary metabolite research predominantly centers on the identification and screening of individual metabolites linked to diseases. However, metabolites operate within intricate networks, engaging in a multitude of synergistic interactions. Meanwhile, the high dimensionality of metabolomics data poses a challenge, revealing limitations in traditional statistical analyses due to constraints in sample size and computing power. Therefore, there is an imperative for an efficient and systematic approach that comprehensively and precisely generalizes the metabolome.
Our study aims to accurately capture the major structures of the plasma metabolome through unsupervised learning, thereby characterizing their genome-wide and phenome-wide profiles and shedding light on their relationships with cardiometabolic diseases. The project duration is 36 months.
Methodologically, our endeavor breaks down complex, high-dimensional metabolomic data into multiple major clusters through unsupervised clustering, utilizing refined scores for cluster characterization. This approach boosts the efficiency of uncovering associations between the metabolome, genomic variation, and diverse human diseases. Clinically, we illuminate epidemiological and genetic ties of metabolic signatures with cardiometabolic diseases, underscoring their potential for precise risk stratification. In addition, genetic variation mining and gene annotation of these metabolic signatures may unveil promising therapeutic targets for cardiometabolic diseases.