Title: Analyzing matched sets of microbiome data
Authors: Yijuan Hu - Emory University (United States) [presenting]
Abstract: Matched data arise frequently in microbiome studies. For example, we may have gut microbiome data pre- and post-treatment from a set of individuals, or longitudinal microbiome samples (e.g., vaginal microbiome samples collected in each trimester of a pregnancy). We present a version of the Linear Decomposition Model (LDM) for analyzing matched datasets. The microbiome characterizing the set is treated as a `nuisance parameter', allowing all effort to focus on the (common) differences within sets. We compare the power of the matched analysis with that of the standard (unmatched) analysis using the LDM.