A probabilistic method for testing and estimating selection differences between populations

Abstract

Human populations around the world encounter various environmental challenges and, consequently, develop genetic adaptations to different selection forces. Identifying the differences in natural selection between populations is critical for understanding the roles of specific genetic variants in evolutionary adaptation. Although numerous methods have been developed to detect genetic loci under recent directional selection, a probabilistic solution for testing and quantifying selection differences between populations is lacking. Here we report the development of a probabilistic method for testing and estimating selection differences between populations. By use of a probabilistic model of genetic drift and selection, we showed that logarithm odds ratios of allele frequencies provide estimates of the differences in selection coefficients between populations. The estimates approximate a normal distribution, and variance can be estimated using genome-wide variants. This allows us to quantify differences in selection coefficients and to determine the confidence intervals of the estimate. Our work also revealed the link between genetic association testing and hypothesis testing of selection differences. It therefore supplies a solution for hypothesis testing of selection differences. This method was applied to a genome-wide data analysis of Han and Tibetan populations. The results confirmed that both the EPAS1 and EGLN1 genes are under statistically different selection in Han and Tibetan populations. We further estimated differences in the selection coefficients for genetic variants involved in melanin formation and determined their confidence intervals between continental population groups. Application of the method to empirical data demonstrated the outstanding capability of this novel approach for testing and quantifying differences in natural selection.

Fast Principal-Component Analysis Reveals Convergent Evolution of ADH1B in Europe and East Asia

“…90 Our work demonstrates the advantages of comparing closely related populations in very large sample sizes to detect subtle signals of selection, whereas very recent studies applying related methods to smaller sample sizes detected genome-wide significant signals only at previously known loci. 91,92 In particular, we detected genome-wide significant evidence of selection in Europeans at ADH1B, which was previously reported to be under selection in East Asian populations 14,55,57,58 using REHH 61 (which can detect only rela-A B C Figure 3. Power of PC-Based Selection Statistic The allele frequency difference at selected SNPs was varied between two populations separated by varying F ST .…”

Fast principal components analysis reveals convergent evolution of ADH1B gene in Europe and East Asia

“…In contrast to previous work on detecting selection using PCs 63,80 or using the spatial ancestry analysis (SPA) method 81 , our statistic is able to detect signals at genome-wide significance, a key consideration in genome scans for selection 82 . Our work demonstrates the advantages of comparing closely related populations in very large sample sizes to detect novel loci, whereas very recent studies applying related methods to smaller sample sizes detected genome-wide significant signals only at previously known loci 83,84 . In particular, we detected genome-wide significant evidence of selection in Europeans at the ADH1B locus, which was previously reported to be under selection in East Asian populations 14,46,48,49 using REHH 53 (which can only detect relatively recent signals and does not work on standing variation 3 ).…”

SeleDiff: A fast and scalable tool for estimating and testing selection differences between populations

“…SeleDiff implements a probabilistic method from our previous study (He et al, 2015). In this approach, we introduced logarithm odds ratios of allele frequencies to measure differences in selective pressures.…”