Online Program Home
My Program

Abstract Details

Activity Number: 615
Type: Contributed
Date/Time: Wednesday, August 3, 2016 : 2:00 PM to 3:50 PM
Sponsor: ENAR
Abstract #319530 View Presentation
Title: Conditional Inference for the Kernel Association Tests
Author(s): Kai Wang*
Companies: University of Iowa
Keywords: kernel machine regression ; genetic association ; permutation ; SKAT ; asymptotic distribution

The kernel association test (KAT) is popular in the analysis of biological data for its ability to combine weak effects potentially of opposite direction. Examples of its applications include human genetic and microbiome association studies. However its (unconditional) asymptotic p-value is conservative when sample size is small, especially for the important case of dichotomous traits. The permutation test, the de facto best approximation to the finite sample conditional inference, is thought to be time-consuming and infeasible. Based on a result of Strasser and Weber (1999), we introduce the conditional asymptotic distribution for the KAT. This distribution provides an improved approximation to the permutation distribution than the small sample distributions previously proposed even when sample size is small. It is also robust to violations of distribution assumptions on the traits. Furthermore, given the simple form of KAT, we investigate and demonstrate the feasibility of permutation for small sample size. The usefulness of these methods are demonstrated via extensive simulation studies using real genotype data and an analysis of genetic data from OHTS.

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

Back to the full JSM 2016 program

Copyright © American Statistical Association