Cormotif: a Hierarchical Mixture Model Framework for Integrating Heterogeneous Genomic Data

• Activity Number: 169
• Type: Contributed
• Date/Time: Monday, August 4, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Biometrics Section
• Author(s): Hongkai Ji*+ and Yingying Wei
• Companies: Johns Hopkins Bloomberg School of Public Health and Johns Hopkins Bloomberg School of Public Health
• Keywords: Bayes hierarchical model ; Correlation motif ; EM algorithm ; Genomics ; Data integration

Abstract

When data from multiple related genomic studies are available, separately analyzing each study is not an ideal strategy as it may fail to detect consistent but relatively weak signals in multiple studies. Jointly modeling all data allows one to borrow information across studies to improve the analysis. However, a simple concordance model, in which each locus is assumed to have signal in either all studies or none of the studies, is incapable of handling study-specific signals. In contrast, a model that naively enumerates and analyzes all possible signal presence and absence patterns across all studies can deal with study-specificity and allow information pooling, but the complexity of its parameter space grows exponentially as the number of studies increases. Here we propose a correlation motif approach to address this dilemma. This approach automatically searches for a small number of latent probability vectors called correlation motifs to capture the major correlation patterns among multiple studies. The motifs provide the basis for sharing information among studies. We demonstrate that the approach can be used to improve differential gene expression analysis in gene expression d