From Scientific American:
Genetic mutations that enhance disease resistance or boost fitness in a particular climate have been positively selected over the course of human evolution. But current statistical methods to find these beneficial mutations, or variants, have only been able to home in on areas spanning several genes, which may cover a variety of other functions, as well. And within these broad swaths of the human genome, there are a number of nonselected, or neutral, variants that also get preferentially inherited because they are on the same chromosome as the selected variant. “It's basically that a whole haystack of [mutations] gets pulled up when selection occurs and then you're trying to find which one was the driver—the needle,” says Pardis Sabeti, an evolutionary geneticist at Harvard University's Department of Organismic and Evolutionary Biology.
To narrow down the culprits for positive selection, Sabeti and her team of researchers developed an approach that combines statistical methods that differ in their specificity for selected variants into one powerful tool. Using the composite tool, the group analyzed mutations in regions from different chromosomes spanning hundreds of kilobases, or hundreds of thousands of nucleotides. The mutations occurred both inside of genes or in the parts of the genome that do not encode genes. Although the team optimized their method for selection that has taken place in the past 30,000 years, Sabeti says that with some tweaking, the approach could stretch back to the point when human populations began migrating out of Africa and diverging from each other 50,000 to 70,000 years ago. The first study using the team's composite method was published January 7 in Science. Using this technique, the scientists could predict an area of the genome as narrow as a single gene, rather than several, that had likely been positively selected. For example, they found a single gene involved in eye color or skin pigmentation that was likely to be selected for from a region containing five genes.
More here.