Hour Talks

Efforts to detect targets of recent positive selection (“selective sweeps”) in the genome frequently involve an initial screen of a region of the genome at a low density of markers (either SNPs or microsatellites), followed by a denser screen in the region flanking markers that show reduced or skewed variation relative to that predicted by an equilibrium neutral model or relative to “background” variation in the genome. Direct sequencing is often then used to attempt to localize the gene/variation that was the target of the selective sweep. I will discuss how non-equilibrium demography seriously complicates the effort to initially detect departures due to natural selection, as well as subsequent efforts to localize the selective target. I will discuss a Goodness of Fit test that, when combined with Kim and Stephan’s composite likelihood method for detecting and localizing sweeps, does reasonably well at distinguishing positive selection from many demographic perturbations. I will also provide examples from our relatively dense microsatellite screen of 800 kb of the X-chromosome in Drosophila melanogaster for both African and non-African population samples with follow-up sequencing that reveal departures from an equilibrium neutral model due apparently to positive selection, to demography alone, and to demographic amplification of an ancestral sweep. Finally, I will discuss our recent analysis of experimental strategies for localizing the targets of selective sweeps, and show that partial sequencing can lead to biased maximum likelihood estimates of selection parameters and reduced rejection rates. For common sample sizes and sampling strategies, the estimate of the target of selection has a very large confidence interval, and the strength of selection is often severely underestimated. A sequencing approach that leads to more accurate estimates will be discussed.