Mostly Biomathematics Lunchtime Seminar

---

What genes and variants do genetic association studies discover?

Speaker: Hakhamanesh Mostafavi, NYU School of Medicine

Location: Warren Weaver Hall 1314

Date: Tuesday, April 23, 2024, 12:45 p.m.

Synopsis:

Genome-wide association studies (GWAS) have uncovered thousands of genetic variants linked to a broad spectrum of human complex traits and diseases. However, relating these discoveries to underlying trait biology remains a significant challenge. Using population genetics theory and data analysis, we demonstrate that GWAS prioritizes variants not solely by their effect size on the studied trait but rather by their specificity to that trait. This is due to the stronger influence of natural selection on pleiotropic variants (i.e., variants that affect multiple traits), resulting in lower statistical power to detect such variants. We further show that GWAS signals partly originate from context-specific regulatory variants acting on highly pleiotropic and selectively constrained genes. These results help explain why GWAS can effectively identify top trait-relevant cell types, while suggesting that it may not be an optimal tool for pinpointing trait-specific genes. Conversely, we show that association studies based on the burden of rare protein-coding variants are expected to prioritize genes based on their trait specificity, further explaining why the rare-variant burden heritability is much less polygenic than common-variant heritability. Taken together, our results suggest new avenues for gaining biological insights from association studies.