New statistical framework allows researchers to better analyze rare genetic variation

June 12, 2014 – Andrew Allen, PhD and colleagues publish new statistical approach for analyzing genomics in the American Journal of Human Genetics.

In a potentially important step for human genetics research, investigators from Duke University, Emory University, and the Centers for Disease Control and Prevention recently released a new statistical framework that has far more power to assess the role of rare genetic variation in human disease.andrew-allen-archive

The approach is detailed in the June 5 issue of the American Journal of Human Genetics. According to senior author Andrew Allen, PhD, (pictured right) this new framework is far more powerful than previous methods for interrogating rare variation because it draws in additional information from large population control samples.

Using these control samples allows researchers to better target their statistical assessment. If a child has an inherited risk variant (a genetic mutation that increases the child’s chance for disease), then one or both of the parents also has to have it. Therefore, there is an enrichment in the parents for these bad variants relative to the general population. Making this comparison allows statisticians to focus on the important variants and reduces the “noise” introduced by neutral variants.

“One of the major issues is that much of the rare variation you find is completely harmless, but some of it can lead to disease,” said Allen. “So, being able to distinguish between those different kinds of rare variation allows you to boost the power of your statistical assessment.”

Allen and his colleagues expect that this new statistical framework will soon make it possible to extract more information from existing samples or to have reasonable power in assessing diseases where large samples are difficult to obtain. For example, in their paper, they apply their approach to data from patients with epileptic encephalopathy, a rare and devastating seizure disorder. They show that their procedure has reasonable power to detect interesting effects while existing approaches have almost no power. They anticipate that their approach will translate into numerous findings across a range of diseases where rare variation plays a role.