Group Association Test Using Hidden Markov Model

• Activity Number: 653
• Type: Contributed
• Date/Time: Thursday, August 13, 2015 : 8:30 AM to 10:20 AM
• Sponsor: Biometrics Section
• Author(s): Yichen Cheng* and James Y. Dai and Charles Kooperberg
• Companies: Fred Hutchinson Cancer Research Center and Fred Hutchinson Cancer Research Center and Fred Hutchinson Cancer Research Center
• Keywords: genome-wide association study ; finite mixture model ; modified likelihood ratio test

Abstract

In the genomic era, group association tests are of great interest. Due to the overwhelming number of individual genomic features, the power of testing for association of a single genomic feature at a time is often very small. Many methods have been proposed to test association of a trait with a group of features within a functional unit, e.g. gene, yet few of these methods account for the fact that generally a substantial proportion of the features are not associated with the trait. We propose to model the association for each feature in the group as a mixture of no association or a constant non-zero association to explicitly account for this fact. The feature-level associations are estimated by generalized linear models; then these estimated associations are modeled by a hidden Markov chain. To test for the global association, we use a modified likelihood ratio test based on an independence log-likelihood with additional penalty term, and derive the asymptotic distribution under the null. Furthermore, we obtain the posterior probability of association for each feature, which provides evidence of feature-level association and will be useful for potential follow up studies.