Selection of Causal Rare Variants in Sequencing Studies

• Activity Number: 288
• Type: Topic Contributed
• Date/Time: Tuesday, August 2, 2011 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistical Computing
• Author(s): Lin Li*+ and Xihong Lin
• Companies: Harvard University and Harvard School of Public Health
• Keywords: variable selection ; sequencing data ; penalized regression

Abstract

There has been increasing interests in studying rare variants and their role underlying human complex diseases, as they may contribute to the genetic component in disease susceptibility that is unexplained by common variants. The advances of re-sequencing methods have made such studies possible, and efforts are taken in searching for regions enriched of causal variants, both rare and common. An important step that follows is to identify individual causal variants from these regions. Naturally variable selection can be applied, but it is challenging as causal rare variants tend to be under-powered to be selected. We propose a weighted penalized regression method for variable selection favoring rare variants. The method is applied to both continuous and binary traits, and various weighting schemes are evaluated. Simulations show that our weighted method is more powerful than unweighted ones in identifying rare causal variants, while taking into account common ones. A real dataset is also studied using our method.