soft sweep

Schrider DR & Kern AD 2017 Soft sweeps are the dominant mode of adaptation in the human genome. Mol Biol Evol 34:1863–1877.

  • soft sweeps are widespread and account for the vast majority of recent human adaptation
  • the search for selective sweeps has shed light into the recent evolutionary histories of natural populations, and has shown a pervasive impact of adaptive evolution on polymorphism in some species such as Drosophila melanogaster
  • in humans, the picture remains less clear
  • some recent studies have suggested that the impact of adaptation on patterns of variation genome-wide is quite limited
  • conversely, Enard et al. (2014) argue that the genome-wide reduction in diversity around substitutions is driven in part by positive selection
  • genetic hitchhiking effects may be muted by human demographic history
  • many human populations appear to have experienced bottlenecks and/or recent growth
  • selection may act on previously segregating neutral or weakly deleterious variants
  • selection on standing variation will produce qualitatively different skews in linkage disequilibrium and allele frequencies, along with a shallower valley in diversity
  • if selection typically proceeds through soft sweeps, as may be the case in Drosophila (Garud et al. 2015), then many sweeps may have been missed by previous scans that were designed to detect signatures produced under a hard sweep model
  • we apply S/HIC to uncover hard and soft sweeps in six population samples from the 1000 Genomes Project (Auton et al. 2015), thereby performing the most comprehensive investigation of completed selective sweeps in humans to date
  • patterns of polymorphism across much of the human genome may be affected by linked positive selection—primarily soft sweeps
  • the mode of selection differs substantially across populations, with non-African populations adapting via hard sweeps to a much greater extent than African populations
  • positive selection on interacting gene pairs
  • we found a significant enrichment for PPIs in hard sweeps
  • this was only significant in for non-African populations
  • for transcription factor–target interactions, we observe no overrepresentation of soft sweeps, but a significant enrichment of hard sweeps in non-African populations
  • there were no populations exhibiting an overrepresentation of pairs of genes with genetic interactions and experiencing sweeps of either type
  • numerous genome-wide scans for selection have been conducted using differing methodologies
  • the majority of these studies searched primarily for partial selective sweeps—the signature of a beneficial mutation currently sweeping through a population
  • these sweeps can reveal the targets of ongoing adaptation in human populations
  • the sojourn of an adaptive mutation to fixation should be rapid
  • e.g., on the order of 400 generations, assuming N = 104 and a moderately strong selection coefficient of s = 0.05, and 4,000 generations for s = 0.005
  • the success of efforts to detect ongoing selection implies the presence of a larger number of recently completed sweeps
  • we have therefore focused on completed sweeps in order to complement previous studies
  • we have recently introduced (S/HIC; Schrider and Kern 2016) for finding completed selective sweeps
  • the majority of our candidate sweeps resemble soft sweeps on standing variation
  • adaptation in humans may not be mutation-limited
  • this may be surprising given the apparently small effective population size and low nucleotide diversity levels in humans
  • if the mutational target for the trait to be selected on is fairly large, then the probability of a population harboring a mutation affecting that trait may be appreciable
  • there is a significant increase in the proportion of putative hard sweeps in non-African populations relative to African populations
  • this is consistent with theoretical expectations
  • larger populations have more standing variation for selection to act on
  • soft sweeps may also occur via recurrent mutation to the adaptive allele
  • there are some qualitative differences between these two models of soft sweep
  • these are fairly subtle in comparison to the differences between the other models we consider
  • the putatively neutrally evolving regions found in this study can be obtained from our raw classification output
  • the phenomenon of deleterious alleles hitchhiking along with strongly beneficial alleles is not restricted to humans
  • a recent study also uncovered evidence that selection during domestication increased the frequency of deleterious polymorphisms in dogs
  • several machine learning methods have been devised to detect selective sweeps
  • they tend to substantially outperform more traditional approaches
  • our scan leveraged a method that performs very well in detecting both hard and soft sweeps
  • it was not trained to detect cases of polygenic selection
  • it could very well be that an even larger portion of genetic variation is influenced by natural selection and its linked effects throughout the genome
  • the S/HIC machine learning approach leverages spatial patterns (along a genome) of a variety of population genetic summary statistics to classify genomic windows as being the target of a completed hard sweep (hard), being closely linked to a hard sweep (hard-linked), a completed soft sweep (soft), linked to a soft sweep (soft-linked), or evolving neutrally (neutral)