Exploiting selection at linked sites to infer the rate and strength of adaptation

Genomic data encode past evolutionary events and have the potential to reveal the strength, rate and biological drivers of adaptation. However, joint estimation of adaptation rate (α) and adaptation strength remains challenging because evolutionary processes such as demography, linkage and non-neutral polymorphism can confound inference. Here, we exploit the influence of background selection to reduce the fixation rate of weakly beneficial alleles to jointly infer the strength and rate of adaptation. We develop a McDonald–Kreitman-based method to infer adaptation rate and strength, and estimate α = 0.135 in human protein-coding sequences, 72% of which is contributed by weakly adaptive variants. We show that, in this adaptation regime, α is reduced ~25% by linkage genome-wide. Moreover, we show that virus-interacting proteins undergo adaptation that is both stronger and nearly twice as frequent as the genome average (α = 0.224, 56% due to strongly beneficial alleles). Our results suggest that, while most adaptation in human proteins is weakly beneficial, adaptation to viruses is often strongly beneficial. Our method provides a robust framework for estimation of adaptation rate and strength across species.