We also examined LD tagging of causal variants by GWAS selected variants within our simulated European and African populations. Each causal variant’s LD score was calculated by summing up the LD r2 between that causal variant and every GWAS tag variant within ± 1,000 kb. The LD scores calculated in European and African ancestry populations were highly correlated (Pearson’s r > 0.7) for the GWAS and fixed-effects meta-analysis selected variants. Variants selected from a fixed-effects meta-analysis had the highest LD score correlation between populations, as expected given that the variants reached significance in both populations and therefore were more common with similar LD patterns (Figure 4B). Since LD score correlation did not vary largely between simulations, we examined the raw LD scores for a single simulation in order to illustrate differences in LD score magnitude not captured by the Pearson’s correlation.
