Event
Ting-Huei Chen, PhD, University of Laval
Tuesday, February 27, 2018 15:30to16:30
Purvis Hall
Room 24, 1020 avenue des Pins Ouest, Montreal, QC, H3A 1A2, CA
A Comprehensive Statistical Framework For Building Polygenic Risk Prediction Models Based On Summary Statistics Of Genome-Wide Association Studies.
Ting-Huei Chen is an assistant professor in the Department of Mathematics and Statistics at University Laval. She focuses on the development of statistical methods for analysis of genetic data. She completed her Ph.D. in Biostatistics at the University of North Carolina at Chapel Hill in 2014 and spent one year as a postdoctoral fellow at the Biostatistics Branch of the National Cancer Institute. https://www.mat.ulaval.ca/departement-et-professeurs/direction-personnel...
Large-scale genome-wide association (GWAS) studies provide opportunities for developing genetic risk prediction models that have the potential to improve disease prevention, intervention or treatment. The key step is to develop polygenic risk score (PRS) models with high predictive performance for a given disease, which typically requires a large training data set for selecting truly associated SNPs and estimating effect sizes accurately. Here, we develop a comprehensive statistical framework, SummaryLasso, to fit regularized regression models based on GWAS summary statistics to develop PRS for both quantitative and binary traits. SummaryLasso is flexible to incorporate information of multiple functional annotations and genetically related traits to further improve the performance of PRS. Extensive simulations show that SummaryLasso performs equally well or better than existing PRS methods when no functional annotation or pleiotropy is incorporated. When functional annotation data and pleiotropy are informative, SummaryLasso substantially outperformed existing PRS methods in simulations. Finally, we compared the performance of different PRS methods on large-scale GWAS of type 2 diabetes (PRS). While the standard PRS had a prediction at the observational scale, SummaryLasso had when incorporating three functional annotations and further improved to when modelling 16 traits that are genetically related with T2D.
Large-scale genome-wide association (GWAS) studies provide opportunities for developing genetic risk prediction models that have the potential to improve disease prevention, intervention or treatment. The key step is to develop polygenic risk score (PRS) models with high predictive performance for a given disease, which typically requires a large training data set for selecting truly associated SNPs and estimating effect sizes accurately. Here, we develop a comprehensive statistical framework, SummaryLasso, to fit regularized regression models based on GWAS summary statistics to develop PRS for both quantitative and binary traits. SummaryLasso is flexible to incorporate information of multiple functional annotations and genetically related traits to further improve the performance of PRS. Extensive simulations show that SummaryLasso performs equally well or better than existing PRS methods when no functional annotation or pleiotropy is incorporated. When functional annotation data and pleiotropy are informative, SummaryLasso substantially outperformed existing PRS methods in simulations. Finally, we compared the performance of different PRS methods on large-scale GWAS of type 2 diabetes (PRS). While the standard PRS had a prediction at the observational scale, SummaryLasso had when incorporating three functional annotations and further improved to when modelling 16 traits that are genetically related with T2D.