Fig. 3

Empirical patterns depend on whether disease-associated alleles are ancestral or derived. a Mean derived allele frequencies of non-disease SNPs from whole genome sequencing and genotyping arrays. 1000 Genomes Project data are shown. b Joint SFS of published GWAS loci. Ancestral risk alleles are labeled red and derived risk alleles are labeled blue. c The frequencies of ancestral risk alleles are higher in Africa (+ 9.51% on average), and the frequencies of derived risk alleles are lower in Africa (− 5.40% on average). Dashed lines indicate mean values. d Continental differences in risk allele frequencies are minimal for young SNPs. Disease-associated loci are binned by DAF and whether risk alleles are ancestral or derived