Evaluating methods for estimating the proportion of adaptive amino acid substitutions

A long-standing debate in molecular evolution concerns the role of adaptation in shaping divergence between species. A number of approaches have been developed to estimate the proportion of amino acid substitutions between species () that are driven by adaptive natural selection. These methods vary in the type of data they use and in the modeling strategies they employ in their inference. In this study, we evaluate the accuracy of nine different methods for estimating , using data simulated in the presence of linked selection. We find that methods that model the distribution of fitness effect (DFE) of both deleterious (as a gamma distribution) and beneficial mutations (as a gamma or exponential distribution) are the most accurate. We applied these methods to whole-genome data, finding that the most accurate methods gave average values of =0.25 in Arabidopsis thaliana, 0.5 in Drosophila melanogaster, and 0.1 in Homo sapiens. We also applied these methods to analyze subsets of tissue-specific genes in A. thaliana that are believed to be under different selective pressures and on genes found on the X vs. autosomes in D. melanogaster. We find estimates of to be higher in the seeds than in other specialized organs, supporting inferences of conflict-driven adaptive evolution in genes expressed in the seed; we also find to be higher on the X chromosome, supporting previous inferences of faster-X evolution. Overall, our results suggest that there are multiple methods that provide accurate estimates of , providing a guide for future estimates of adaptive evolution.