Nucleotide diversity (π) varied considerably among genes, ranging from ∼0 to 1.319% per bp (mean = 0.042%; Fig. 1B). Mean π in AAs (0.047%) was significantly higher (P < 10−15, paired t test) than π in EAs (0.035%), and π per gene was modestly correlated (r2 = 63%; P value < 10−15) between AAs and EAs (fig. S15). Rare variants account for 4% of total diversity, more than any other MAF bin (of width 0.5%) in both EAs and AAs (Fig. 1C). Rare and low-frequency SNVs comprise ∼13 and 20% of total diversity in the EA and AA samples, respectively (Fig. 1C). In both samples, estimates of π were highest for human lymphocyte antigen (HLA) loci and other genes related to immune function, such as DEFB108B, and olfactory receptors (Fig. 1B). When genes were grouped into functional categories by KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway, estimates of π were highest for pathways related to immune function and olfaction and lowest for pathways involved in basic cellular processes (fig. S16).
