Improving Genetic Risk Prediction by Leveraging Pleiotropy

• Activity Number: 589
• Type: Topic Contributed
• Date/Time: Wednesday, August 7, 2013 : 2:00 PM to 3:50 PM
• Sponsor: ENAR
• Author(s): Cong Li*+ and Jia Kang and Can Yang and Hongyu Zhao
• Companies: Yale University and Merck and Yale University and Yale University
• Keywords: genetic risk prediction ; GWAS ; pleiotropy

Abstract

Although hundreds of genome-wide association studies (GWAS) have been conducted on many hum complex traits in the last 8 years, there is only limited, if any, success in translating these GWAS data into useful genetic prediction models. This is largely due to the presence of numerous genetic variants carrying only small to moderate genetic effects that cannot be easily identified given the limited sample size. Recent studies have shown that different human traits may share common genetic basis. Therefore, an attractive approach to increase our prediction capability is to integrate data of genetically correlated phenotypes. Yet the utility of genetic correlation in genetic prediction is not clearly understood. In this paper, we analyzed a GWAS data of bipolar and related disorders (BARD) and schizophrenia (SC) with a bivariate linear mixed model based method and found that jointly predicting the two phenotypes increased the AUC by 3% and 0.7% for BARD and SC respectively. We also performed comprehensive simulation studies to investigate the utility of genetic correlation in genetic risk prediction. We show that when there exists substantial genetic correlation between two phenotypes