Sweeping Out Selective Sweeps


The low diversity "trough" around human-specific gene substitutions (from Hernandez et al, Science, 2011)

The ultimate genetic detective story is solving the mystery of human evolution. Since it became possible to look at genetic sequences in humans and their primate relatives, geneticists have hunted for the footprints of how humans evolved. But finding the most significant places in the genome that changed since humans and chimpanzees split off from their common ancestor 6 million years ago isn’t easy. With more than 3 billion base pairs in the human genome – which we have only recently been able to sequence in full – geneticists needed to develop strategies to narrow down the potential places of key genetic changes. But what if those strategies were wrong?

That’s the seismic question asked by a paper in Science last week by former University of Chicago post-doctoral researcher Ryan Hernandez, professor of human genetics and evolution & ecology Molly Przeworski, and scientists from England and Israel. Using data from the massive 1000 Genomes Project, the researchers tested the the classic selective sweep model, the predominant strategy used to look for evolutionary footprints for the last 35 years. But with a new, larger dataset, Hernandez and colleagues found little evidence for selective sweeps in the history of our species. Instead, the evolution of humans may have progressed by baby steps, rather than large strides.

Since a 1974 paper on the “hitch-hiking” of favorable genes, geneticists have hunted for selective sweeps in the human genome. As the theory goes, when a beneficial mutation appeared in early humans, it conferred such an immediate, large advantage to its carriers that it quickly “swept” through the entire population. That rapid spread should leave a detectable mark in the human genome – a patch of reduced diversity, where propagation of the gene out-paced the gene-shuffling effect of recombination.

By looking for regions of low diversity in the genome, geneticists were able to identify approximately 2,000 human genes (roughly 10 percent of the total) that showed evidence of selective sweeps. Such a high frequency suggested that selective sweeps were the leading mechanism by which humans evolved away from their primate relatives, as well as how different modern populations (i.e. African, Asian, or European origin) evolved smaller differences in the last 100,000 years.

“The selective sweep model was introduced in 1974 and has pretty much been the central model ever since,” Przeworski said. “It is fair to say that it is the model behind almost every scan for selection done to date, in humans or in other organisms.”

But in order for the selective sweep model to hold up, the patches of low diversity used to find them would have to only occur around genes with meaningful changes in humans. Using 179 of the genomes collected for the 1000 Genomes Project – a much larger set of data then ever available before – Hernandez, Przeworski, and colleagues compared the diversity surrounding two groups of human-specific genetic changes, where a new base pair was introduced into the gene code. In the first group, the new base-pair changed the code such that a new amino acid was put into the protein encoded by the human gene, presumably changing its function. In the second group, the new base-pair was “synonymous,” encoding for the same amino acid and leaving the encoded protein unperturbed.

But when the two groups were compared, the “trough” of low diversity was equal for both changes with functional effects and changes with no effect at all. The result suggests that low diversity is false evidence for a selective sweep, and that those sweeps are likely to have been only occasional contributors in the story of human evolution.

“Phenotypic variation in humans isn’t as simple as we thought it would be,” said Hernandez, now an assistant professor of computational and systems biology at the University of California, San Francisco. “The idea that human adaptation might proceed by single changes at the amino acid level is quite a nice idea, and it’s great that we have a few concrete examples of where that occurred, but it’s too simplistic a view.”

The point was reinforced by another comparison, of genomes from different geographic populations. If selective sweeps were a common occurrence in human evolution, they would have locked in place different gene variants between East Asian, African, and European populations that have been living isolated in different parts of the world for most of the last 100,000 years. However, comparing representatives of those genetic backgrounds from the 1000 Genome Project found only subtle differences, such as 60 percent of a particular gene variant in the European population and 40 percent of that variant in genomes of African origin.

So while it might be satisfying to think of human evolution as being marked by the sudden appearance and spread of new genes like pop-music phenoms, the truth is probably more glacial. Though a few important single-gene changes probably did occur, the evolution of humans was probably more often propelled by subtle changes in multiple genes, coming together to produce human phenotypes such as increased intelligence, upright posture, and complex spoken language. A model of human evolution progressing by the accumulation of small tweaks fits with what geneticists are learning about human disease in genome-wide association studies, where common diseases are rarely explained by single genes.

“Our findings suggest that recent human adaptation has not taken place through the arrival and spread of single changes of large effect, but through shifts of frequency in many places of the genome,” said Przeworski. “It suggests that human adaptation, like most common human diseases, has a complex genetic architecture.”


Hernandez RD, Kelley JL, Elyashiv E, Melton SC, Auton A, McVean G, 1000 Genomes Project, Sella G, & Przeworski M (2011). Classic Selective Sweeps Were Rare in Recent Human Evolution. Science (New York, N.Y.), 331 (6019), 920-924 PMID: 21330547

About Rob Mitchum (526 Articles)

Rob Mitchum is communications manager at the Computation Institute, a joint initiative between The University of Chicago and Argonne National Laboratory.

%d bloggers like this: