B0699
Title: Statistical inference for joint factor regression with applications to integrating multi-view microbiome data
Authors: Jing Ma - Fred Hutchinson Cancer Center (United States) [presenting]
Abstract: Mechanistic understanding of the microbiome requires identifying co-regulating microbial markers (e.g. taxa, metabolites, etc.) that are associated with host health outcomes. A joint factor regression model for analysis of multi-view microbiome data is discussed (e.g. metagenomics, metabolomics, etc.). Unlike existing methods that focus only on variable selection, our method identifies a multivariate association between the outcome and the latent factors common to all omics layers. It also enables hypothesis testing of specific variables underlying the multivariate association. We will illustrate the merit of the proposed method using an analysis of metagenomic and metabolomic data from the Study of Latinos.