In their article “Classic Selective Sweeps were Rare in Recent Human Evolution” Hernandez et al. 2011 were interested in the modes of natural selection that shaped human adaptations. Up to date, most studies suggest that the principal mode of adaptation is due to positive selection. Therefore, a beneficial mutation appears in a population and is getting rapidly fixed. The decrease in neutral diversity in the linked sites results in the occurrence of a ‘classic selective sweep’. Hernandez et al. 2011 were questioning whether it could be possible that not only selective sweeps but also other types of selection could have been involved in human adaptation.
Resequencing data for 179 human genomes from “three” populations (African, Chinese/Japanese and European) was investigated. They assessed average diversity levels as a function of genetic distance from the nearest exon and the nearest conserved non-coding region. If functional changes in amino acids would result in a classic selective sweep, the diversity level of non-synonymous substitutions would decrease in comparison to synonymous substitutions. This pattern has already been confirmed in Drosophila simulans. Interestingly the authors revealed a decrease in both, synonymous and non-synonymous substitutions. Hence, they suggest instead strong purifying selection on linked size to explain the pattern. So far it has been believed that synonymous sites evolve neutrally in mammals. But recent studies demonstrate that synonymous sites are important in mRNA stability and for correct splicing. So, the decrease in diversity could maybe also be linked to positive selection?
Moreover, tests for classic sweeps were carried out, by comparing the genetic differentiation of the three populations. An enrichment of highly differentiated single nucleotide polymorphisms (SNPs) between pairs of populations in genic regions has been unravelled. So at least some SNPs might have evolved through the action of positive selection according to Hernandez et al. 2011.
However, tests of highly differentiated alleles at non-synonymous sites, transcription start sites and 5’ or 3’ untranslated regions against the genomic background were almost or not at all significant. This suggests that the differentiated alleles were most probably selected from standing genetic variation. This is supported by the fact that alleles with very high differences in frequencies often segregate in both compared populations and tend to lie on shorter haplotypes than expected from classic sweeps. But maybe there might also be the possibility that ‘neutral sweeps’ could have occurred during evolution. The probability is quite low but when populations expand alleles can get fixed by chance, which is a genetic signature of the ‘founder effect’.
All in all, a lot of the hypotheses that have been suggested remain unanswered, referring to future research. Figures were hard to understand, especially when the legend is not comprehensive. It also took some time to go through the article that was referring a lot of times to the supplementary material (54 pages!). But I really appreciate the effort to give a short and comprehensively written overview for the huge amount of work that has been realized.
Hernandez R.D., Kelley J.L., Elyashiv E., Melton S.C., Auton A., McVean J., 1000 Genomes Project, Sella G., Przeworski M. (2011). Population Genomics of Parallel Adaptation in Threespine Stickleback using Sequenced RAD Tags Nature, 6 DOI: http://www.sciencemag.org/content/331/6019/920.full