Computational framework for detecting positive and negative enrichment for evolutionary signatures on genome-wide association studies (GWASs).

(a) Given the GWAS of a complex trait, we define trait-associated regions by first identifying variants of genome-wide significance and then clumping based on linkage disequilibrium (LD; e.g., r2>0.9). For each region, we identify the maximum value of an evolutionary measure of interest. (b) For each trait-associated region, we identify 5,000 randomly selected genomic regions (“matched regions”) that have similar minor allele frequency, linkage disequilibrium, and gene proximity patterns (Methods). (c) Across the trait-associated regions and their matched random genomic regions, we calculate a summary statistic. To illustrate our approach, we take the mean of the evolutionary measure to generate an (d) empirical background distribution and (e) calculate enrichment by comparing the mean observed evolutionary measure to the mean of the matched background distribution. We divide by the standard deviation of the evolutionary measure across the genome to standardize the enrichment. However, any summary statistic of interest could be used.