McDonald–Kreitman test, neutrality index, and Direction of Selection

We assessed the possible fixation of variants in the Homo sapiens population by first calculating the relative ratio of nonsynonymous to synonymous polymorphism (p_N/p_S) from the 1000 Genomes VCF for all SNPs, for SNPs with a minor allele frequency (MAF) <1% and <5%. SNPs were annotated with ANNOVAR across 1000 Genomes Project (ALL+5 ethnicity groups), ESP6500 (ALL+2 ethnicity groups), ExAC (ALL+7 ethnicity groups), and CG46 (for more details, see http://annovar.openbioinformatics.org/en/latest/user-guide/filter/#popfreqmax-and-popfreqall-annotations). The polymorphism ratio (p_N/p_S) allowed us to take into account the constraint on nonsynonymous sites and thus increase the power of detecting positive selection (Salvador-Martínez et al. 2018). We indeed normalized the divergence ratio (d_N/d_S) using the McDonald–Kreitman (MK) test, that is, calculating the neutrality index (NI) as the ratio of raw p_N/p_S and d_N/d_S values (McDonald and Kreitman 1991). We considered the PSG to be fixed in the population when NI < 1. We also confirmed with a new statistic for evolutionary measure: the Direction of Selection (DoS) = D_n/(D_n + D_s) − P_n/(P_n + P_s) (Stoletzki and Eyre-Walker 2011) that all divergent genes with NI < 0 had a DoS < 0 (Supplemental Fig. S8).