Shifts in the intensity of purifying selection: an analysis of genome-wide polymorphism data from two closely related yeast species

Abstract

How much does the intensity of purifying selection vary among populations and species? How uniform are the shifts in selective pressures across the genome? To address these questions, we took advantage of a recent, whole-genome polymorphism data set from two closely related species of yeast, Saccharomyces cerevisiae and S. paradoxus, paying close attention to the population structure within these species. We found that the average intensity of purifying selection on amino acid sites varies markedly among populations and between species. As expected in the presence of extensive weakly deleterious mutations, the effect of purifying selection is substantially weaker on single nucleotide polymorphisms (SNPs) segregating within populations than on SNPs fixed between population samples. Also in accordance with a Nearly Neutral model, the variation in the intensity of purifying selection across populations corresponds almost perfectly to simple measures of their effective size. As a first step toward understanding the processes generating these patterns, we sought to tease apart the relative importance of systematic, genome-wide changes in the efficacy of selection, such as those expected from demographic processes and of gene-specific changes, which may be expected after a shift in selective pressures. For that purpose, we developed a new model for the evolution of purifying selection between populations and inferred its parameters from the genome-wide data using a likelihood approach. We found that most, but not all changes seem to be explained by systematic shifts in the efficacy of selection. One population, the sake-derived strains of S. cerevisiae, however, also shows extensive gene-specific changes, plausibly associated with domestication. These findings have important implications for our understanding of purifying selection as well as for estimates of the rate of molecular adaptation in yeast and in other species.

Figure 1.

The relationship between the intensity of purifying selection and the effective population size. The intensity of purifying selection is measured as the genome-wide estimate of f_π; the two proxies used for the effective population size are the genome-wide average synonymous heterozygosity values, (A) and (B), where is the average synonymous divergence (see text). Central 95th percentiles for the estimates, represented by the horizontal and vertical bars, were estimated by bootstrapping over genes. The dashed line in B shows the expected relationship between f and assuming a gamma-shaped distribution of mutational selective effects with a point mass at s = −∞ (see text and Methods). (C) Using the same distribution, the estimated fractions of effectively neutral, weakly deleterious, and strongly deleterious amino acid mutations in each population.

Figure 2.

Correlations between per-gene estimates of f_ϕ in two populations. (A) The estimates of f_ϕ for each gene in the European samples of S. cerevisiae and S. paradoxus. The dashed lines mark the range of possible values of the parameters. (B) A similar scatter plot generated by simulation, under a model where f for each gene is identical in the two populations.

Figure 3.

Our model for shifts between two populations in the intensity of purifying selection on a gene. See text for details. Subindices 1 and 2 refer to population 1 and 2, respectively. S(f) denotes the stochastic transformation and D(f) the deterministic transformation. Dark gray represents the central 50th percentiles of the probability mass, and light gray represents the central 95th percentiles. The arrows in D and E represent the values of f before and after the transformation.

Figure 4.

Testing the fit of the model for the European populations of S. cerevisiae and S. paradoxus. We compared five summaries of the distribution of observed synonymous polymorphism levels, p_S, and nonsynonymous polymorphism levels, p_N, to what is expected under the model with the ML-parameter estimates (based on 1000 simulated data sets; gray lines are the observed; black bars are the histogram of simulated values). Shown are results for the mean and standard deviation in each population as well as the correlation between the two populations; below each histogram are the P-values associated with the observed values. Similar fits were obtained for the other pairs of populations (Supplemental Table 2).

Figure 5.

Correlation between the intensity of purifying selection on a gene in different populations between species (A) and within species (B). The number of gray squares represents the sample size in each population. (A) The estimated correlation coefficient in the comparison between population samples from S. cerevisiae and S. paradoxus. The gray arrows point to the sister population that was used to call SNPs as fixed (see Methods). (B) The estimated correlation coefficient based on fixed SNPs between two populations, X and Y, and fixed SNPs between the common ancestor of these populations and a third population, Z, from the same species. See text for further details. The central 80th percentiles of the correlation coefficient (in gray) were estimated by parametric bootstrap (with 100 replicates).

Figure 6.

Histograms of P-values across 1727 genes obtained for the McDonald-Kreitman test based on SNP data in S. paradoxus, using two different sampling strategies. P-values for each gene were classified into 20 bins of width 0.05, and smoothed using LOWESS (Cleveland and Devlin 1988). Note that we used a different scale on the y-axes of the two histograms. While near 0, no excess of P-values is observed using SNPs segregating in the European sample (A), such an excess (highlighted by the dashed rectangle) is observed using SNPs from the entire population sample (B). This excess provides evidence for the existence of genes that experienced adaptive evolution, and allows us to estimate a lower bound on their number.