The Interplay of Linkage Disequilibrium Score Regression and Selection
The current method to estimate heritability based on genome-wide association studies (GWAS) summary statistics is to use Linkage Disequilibrium Score Regression (LDSC). Linkage Disequilibrium (LD) describes the association of genetic variants and LDSC makes use of the fact that variants which tag many other variants through high LD are more likely to tag a causal variant (Bulik-Sullivan et al. 2015). In fact, there is a linear relationship between the square of a variants normalized effect size and its LD score. Therefore, LDSC can be used to estimate both heritability and genetic correlations between traits by controlling for the effects of stratification in GWAS cohorts. A key assumption is that LDSC uses the independence of LD score and population differentiation. However, this assumption is violated when the model includes selection. This project aims to investigate the robustness of LDSC in scenarios involving varying types of selection, such as positive selection, background selection, and selective sweeps.
Message to Sponsor
- Major: Data Science
- Sponsor: Rose Hills Experience
- Mentor: Rasmus Nielsen