Ridge Regression To Accommodate LD in WGA Studies

• Activity Number: 332
• Type: Contributed
• Date/Time: Tuesday, July 31, 2007 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Nathalie Malo*+ and Jennifer Wessel and Nicholas J. Schork
• Companies: University of California, San Diego and University of California, San Diego and University of California, San Diego
• Keywords: association analysis ; epistasis ; genetic variation ; genome-wide scans ; target SNPs

Abstract

The use of whole genome association (WGA) studies for the identification of genes and genetic variations that influence common, complex diseases such as hypertension, cancer, and depression will continue to grow as cost-effective high-throughput genotyping technologies are developed. As a result, appropriately flexible yet robust data analysis strategies for analyzing WGA data will be essential. We emphasize the need to accommodate phenomena such as linkage disequilibrium via simple extensions of traditional regression models. We describe the use of regression analysis models for WGA that are very intuitive and flexible. We propose the use of ridge regression, a special case of Bayesian regression, to account for correlation. We showcase the utility of the method on previously published WGA data. We also consider limitations of the proposed approach as well as areas for further research.