Chen Liu
Title: A comparison of two extreme-phenotype sampling designs for sequencing studies of disease endophenotypes
Date: Wednesday, April 15th
Time: 9:30am
Location: LIB 2020 & Zoom
Supervised by: Jinko Graham
Abstract: In genetic studies, extreme-phenotype sampling (EPS) is a two-phase design that increases power and reduces sequencing costs by selecting individuals with extreme values of a quantitative trait for sequencing. Quantitative traits are typically endophenotypes, viewed to be more genetically influenced than the disease. Traditional EPS selects individuals with the highest and lowest trait values, while its lesser-known counterpart, extreme-residual sampling (ERS), selects based on residuals after covariate adjustment. Standard analyses dichotomize the trait, discarding valuable detail. We implement an approach that preserves the trait鈥檚 continuous nature to compare the two designs. We apply this approach to a quantitative trait modeled on intraocular pressure, a glaucoma endophenotype, and simulated sequence data from a glaucoma-linked genomic region. Our results indicate that the ERS design more efficiently uses first-phase resources than routine EPS, supporting its adoption as the default study design in this context.