Analysis of Sequencing Studies Under Multivariate Trait-Dependent Sampling

• Activity Number: 303
• Type: Contributed
• Date/Time: Tuesday, August 6, 2013 : 8:30 AM to 10:20 AM
• Sponsor: Biometrics Section
• Author(s): Ran Tao*+ and Danyu Lin and Donglin Zeng
• Companies: Department of Biostatistics, The University of North Carolina and Univ of North Carolina and The University of North Carolina
• Keywords: EM algorithm ; Linear regression ; Nonparametric maximum likelihood estimation ; Outcome-dependent sampling ; Quantitative traits ; Semiparametric inference

Abstract

High-throughput DNA sequencing is a cutting-edge technology for genetic association studies. Currently, it is prohibitively expensive to sequence all subjects in a large cohort. A cost-effective strategy is to preferentially sequence the subjects with the extreme values of a quantitative trait. We consider the situation in which the sampling depends on multiple quantitative traits and possibly also on covariates. Under such outcome-dependent sampling, standard linear regression analysis is invalid and inefficient. We construct a semiparametric likelihood that properly reflects the sampling mechanism. In our formulation, quantitative traits are related to genetic variables and covariates through a multivariate linear regression model while the distributions of genetic variables and covariates are arbitrary. We develop a novel EM algorithm to maximize the likelihood and establish the asymptotical properties of the resulting estimators. Simulation studies demonstrate the superiority of the proposed methods over standard linear regression methods. Data from the ongoing Cohorts for Heart and Aging Research in Genomic Epidemiology Targeted Sequencing Study (CHARGE-TSS) are provided.