Robust Principal Component Analysis For Population Stratification In Genome-Wide Association Studies

• Activity Number: 472
• Type: Contributed
• Date/Time: Wednesday, August 3, 2011 : 8:30 AM to 10:20 AM
• Sponsor: Biopharmaceutical Section
• Author(s): Li Liu*+ and Donghui Zhang
• Companies: sanofi-aventis and sanofi-aventis
• Keywords: population stratification ; genome wide association studies ; robust principal component analysis ; resampling by half means (RHM) ; outlier detection

Abstract

Genome-wide association studies can be useful to understand the diseases of interests or the responsiveness of an individual to certain treatments. In such studies, it is very important to correct for population stratification, which refers to allele frequency differences between cases and controls due to systematic ancestry differences. Population stratification can cause spurious associations if not adjusted properly. Currently, principal component analysis (PCA) has been widely used to adjust for population stratification and proven to be useful in many situations. This approach can be easily applied to thousands of markers, and the correction is specific to a marker's variation across ancestral populations. However, the PCA approach may not be able to properly correct for population stratification in the presence of outliers since PCA is very sensitive to outliers. We propose to use robust principal component analysis combined with clustering to deal with population stratification. This approach can adjust population stratification for both continuous and discrete populations with outliers. Simulation studies were used to demonstrate the usefulness of this approach.