Online Program Home
  My Program

All Times EDT

Abstract Details

Activity Number: 231 - SPEED: SPAAC SESSION I
Type: Topic-Contributed
Date/Time: Wednesday, August 11, 2021 : 10:00 AM to 11:50 AM
Sponsor: Section on Statistics in Epidemiology
Abstract #318352
Title: High-Dimensional Regression of Continuous Secondary Traits Under Extreme Phenotype Sampling
Author(s): Lance Ford* and Chao Xu
Companies: University of Oklahoma Health Sciences Center and University of Oklahoma Health Sciences Center
Keywords: Statistical genetics; Extreme phenotype sampling; High-dimensional regression; Secondary-trait analysis; Simulation studies

Extreme phenotype sampling (EPS) has been shown to improve efficiency of parameter estimation on small samples. Secondary phenotype traits (SPTs) can be evaluated after sampling extremes of the primary phenotype trait (PPT). Maximum likelihood (ML) methods have been developed to evaluate SPTs under EPS. However, we hypothesize that ML methods are unreliable in high-dimensional (HD) regression. Our objective is to determine when ML approaches are unreliable. We will conduct simulations on a constructed dataset of 1,000 individuals that includes 10,000 gene expressions per individual and PPT and SPT information. Individuals whose PPT value falls in the top or bottom 20th percentiles will be included in the analysis. We will then assess the association between the SPT and the gene expressions. The number of gene expressions will be varied to determine when the ML approach is unreliable in terms of bias and efficiency. We anticipate that models with large numbers of gene expressions will perform poorly. This anticipated limitation motivates the development of HD methods to analyze such data. This could lead to improved statistical power and reduced cost in gene association studies.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2021 program