[Selective structure of the gene pool. IV. Estimation from the selection intensity index Rs].

A new approach for investigating the selective structure of the gene pool reflecting the type and intensity of selection is proposed. Selection pressure is estimated on the basis of interpopulation gene diversity with the use of the selection intensity index: RS(i) = NeS(i) = 1/4(1/FST(i)-1/Fe). Distributions of RS(i) in gene pools of indigenous populations from all continents and five subregions of the northeastern Eurasia were examined. It was shown that, of all theoretical distributions, only beta-distributions provide a good approximation of RS(i) estimates. Based on the confidence intervals of RS obtained from beta-distributions, genes can be grouped into the three following classes according to their selective structure: LOWER DIFF, NEUTRAL, and SUPER DIFF. These classes, respectively, include genes subjected mainly to stabilizing selection (RS(i) > 0; LOWER DIFF), genes subjected mainly to differentiating selection (RS(i) < 0; SUPER DIFF), and arbitrarily selectively neutral genes (RS(i) approximately 0; NEUTRAL). Simulation of gene pool sampling (10(6) samples from 50 markers for each gene pool) allowed us to characterize the selective structure by determining markers that fall into the same selective class irrespective of the variant for the sampling process. The selective structure of gene pools from six continents (Europe, Asia, Africa, Australia, America, and southeastern Eurasia) and five subregions of northeastern Eurasia was characterized. It was shown that approximately one-third of genes is subjected to selection irrespective of the hierarchical level of the region. In gene pools of Europe, northeastern Eurasia, and European and Ural subregions, the proportion of genes under stabilizing selection was higher, the proportion of selectively neutral genes, lower. Debatable issues of tests for selective neutrality based on heterogeneity of interpopulation gene diversity are considered. These issues include the effect on FST of the hierarchical population structure, sample size, number of subpopulations, and other factors that shift estimates of gene selective values.