A "selective sweep" is basically a discrete instance of a gene variant's frequency changing due to some form of natural selection (including reproductive partner selection, fertility rates, and death effects). A new study discussed below looked at genetic evidence of such sweeps in 5 populations each in Africa, Europe and Asia over the last 100,000 years which is about 3,500 generations based upon modern genomes from a pre-existing database. The populations studies were as follows:
The accuracy of the method was confirmed, among other reasons, because the blind statistical methods used located the most famous and intense selective sweeps previously known to have occurred. As the body text explains:
We found that the most enriched genomic regions contain multiple iconic examples of positive selection, e.g., the LCT region in northern Europeans, EDAR in all Asian populations and TLR5 in Africa, as well as various other examples including genes associated with lighter skin pigmentation. This indicates that the information underlying our estimations corresponds to genomic signals in line with natural selection.
Africa has had fewer selective sweeps, instances of gene frequency rate changes not caused by mere random drift, in the last 100,000 years than Europe, which has in turn had fewer selective sweeps than Asia. As the body text explains:
We also found a trend toward more sweeps among non-African populations, particularly in Asia. For example, we estimated 68 [CI:46-91] and 165 [CI:119-211] sweeps in average in Africa and Asia respectively when including complete sweeps in our model.
This is, in part, due to the fact that Africa has had fewer and less intense population bottlenecks in this time frame. It is also due, in part, to the fact that humans, who evolved in the first place in Africa, needed fewer evolutionary adaptations to continue to live there than to live elsewhere where different environmental conditions were present.
Most selective sweeps have been limited to particular continents, and not global, but have been continent-wide. As the body text of the paper explains:
We found virtually no overlap between populations from different continents, with some exceptions.
In contrast, the most enriched regions tend to be shared across populations from the same continent. These results indicate that the total number of sweeps in humans should be close to the summation of the mean X computed per continent; i.e., 221 [CI:122-329], a first order approximation neglecting some selection signals shared across continents and considering sweeps are highly shared within continents.
Figure 4C below shows the results graphically with margins of error for Europe and Asia. Below that is a similar figure for five African populations from the Supplemental Materials. The abstract for the paper and its citation follow.
Over the last 100,000 years, humans have spread across the globe and encountered a highly diverse set of environments to which they have had to adapt. Genome-wide scans of selection are powerful to detect selective sweeps. However, because of unknown fractions of undetected sweeps and false discoveries, the numbers of detected sweeps often poorly reflect actual numbers of selective sweeps in populations. The thousands of soft sweeps on standing variation recently evidenced in humans have also been interpreted as a majority of mis-classified neutral regions. In such a context, the extent of human adaptation remains little understood.
We present a new rationale to estimate these actual numbers of sweeps expected over the last 100,000 years (denoted by X) from genome-wide population data, both considering hard sweeps and selective sweeps on standing variation. We implemented an approximate Bayesian computation framework and showed, based on computer simulations, that such a method can properly estimate X. We then jointly estimated the number of selective sweeps, their mean intensity and age in several 1000G African, European and Asian populations.
Our estimations of X, found weakly sensitive to demographic misspecifications, revealed very limited numbers of sweeps regardless the frequency of the selected alleles at the onset of selection and the completion of sweeps. We estimated ∼80 sweeps in average across fifteen 1000G populations when assuming incomplete sweeps only and ∼140 selective sweeps in non-African populations when incorporating complete sweeps in our simulations. The method proposed may help to address controversies on the number of selective sweeps in populations, guiding further genome-wide investigations of recent positive selection.
Guillaume Laval, et al., "A genome-wide Approximate Bayesian Computation approach suggests only limited numbers of soft sweeps in humans over the last 100,000 years" bioRxiv (December 23, 2019) doi: https://doi.org/10.1101/2019.12.22.886234
No comments:
Post a Comment