Precision medicine for atherosclerotic cardiovascular disease: Integrative genomics maps risk loci and AI-predicted functional consequences.
Atherosclerotic cardiovascular disease (ASCVD) is a leading cause of global morbidity and mortality, but its genetic architecture remains incompletely understood. This study aims to uncover novel genetic insights into ASCVD through multivariate genomic analysis and molecular structure predictions.
We analysed genomic data from over 3.8 million individuals across various ASCVD phenotypes, including coronary heart disease, stroke, transient ischemic attack, peripheral artery disease and abdominal aortic aneurysm. Advanced tools such as Genomic Structural Equation Modeling, fine-mapping, FUSION, FOCUS and other functional annotation methods were applied to identify causal single nucleotide polymorphisms associated with ASCVD. Protein structural analysis was performed using AlphaFold3, and AI-driven thermodynamic analysis (ThermoMPNN) assessed the stability and functional consequences of mutations.
Our analysis revealed 347 genome-wide significant variants linked to ASCVD, distributed across 213 loci. Ninety of these variants were not identified in any of the five input GWAS datasets. Upon cross‑referencing with large‑scale external GWAS, 19 of the 90 variants showed no prior association with any cardiovascular or metabolic trait, 15 were previously reported only in risk factor GWAS, and 56 had been reported in direct ASCVD endpoint GWAS. The latter group includes the DCLRE1B rs11552449 missense mutation. Nevertheless, AI‑based structural and thermodynamic analyses revealed that this mutation (H61Y) disrupts DCLRE1B protein stability, increases conformational flexibility, and alters solvent‑accessible surface area-mechanistic insights that have not been previously described.
This study provides a hypothesis‑generating genetic landscape of ASCVD, unveiling novel variants and their molecular impacts. These findings enhance our understanding of ASCVD mechanisms and may offer potential avenues pending experimental validation and targeted therapies in cardiovascular disease.
We analysed genomic data from over 3.8 million individuals across various ASCVD phenotypes, including coronary heart disease, stroke, transient ischemic attack, peripheral artery disease and abdominal aortic aneurysm. Advanced tools such as Genomic Structural Equation Modeling, fine-mapping, FUSION, FOCUS and other functional annotation methods were applied to identify causal single nucleotide polymorphisms associated with ASCVD. Protein structural analysis was performed using AlphaFold3, and AI-driven thermodynamic analysis (ThermoMPNN) assessed the stability and functional consequences of mutations.
Our analysis revealed 347 genome-wide significant variants linked to ASCVD, distributed across 213 loci. Ninety of these variants were not identified in any of the five input GWAS datasets. Upon cross‑referencing with large‑scale external GWAS, 19 of the 90 variants showed no prior association with any cardiovascular or metabolic trait, 15 were previously reported only in risk factor GWAS, and 56 had been reported in direct ASCVD endpoint GWAS. The latter group includes the DCLRE1B rs11552449 missense mutation. Nevertheless, AI‑based structural and thermodynamic analyses revealed that this mutation (H61Y) disrupts DCLRE1B protein stability, increases conformational flexibility, and alters solvent‑accessible surface area-mechanistic insights that have not been previously described.
This study provides a hypothesis‑generating genetic landscape of ASCVD, unveiling novel variants and their molecular impacts. These findings enhance our understanding of ASCVD mechanisms and may offer potential avenues pending experimental validation and targeted therapies in cardiovascular disease.