Masters candidate in Biostatistics, Yuanyuan Ji, will present:
“Application of Bayesian Hierarchical Structured Variable Selection Method in Multiple Traits GWAS Summary Statistics”
Plan B Adviser: Lin Zhang
Abstract: Genome-Wide Association Studies (GWAS) have evolved over the last ten years into a powerful tool for the investigation of human disease genetic variation. It has discovered hundreds of genetic variants associated with disease and quantitative traits. However, published GWAS studies are almost univariate and only test associations between one trait and one genetic variant at a time. In the study of complex disease, correlated traits are often measured as risk factors and influenced by genetic variants simultaneously. Hence, testing for genetic association with multiple traits becomes important due to its potentiality of boosting statistical power. Recent studies showed that serum concentrations of low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, and triglycerides are among the most important risk factors for coronary artery disease. To investigate the association of genetic variants and serum lipids, genome from over 100,000 European individuals were screened and its GWAS summary statistics was made publicly available. In this study we explored the performance of a Bayesian hierarchical structured variable selection (HSVS) method in association analysis of multiple traits using this lipids summary statistics. This approach utilized a prior for both group selection and within group shrinkage in the presence of natural grouping structure. After analyzing associations for a total of 2.7M SNPs, HSVS approach captured the majority SNPs that were identified as significant from other commonly used frequentist methods. Some of significant SNPs and loci we identified were also supported by previous lipids GWAS studies.
Refreshments will be served prior to the presentation.