Title: Logistic normal multinominal biclustering mixture model for microbiome count data
Authors: Wangshu Tu - Binghamton University (United States) [presenting]
Sanjeena Dang - University of Guelph (Canada)
Yuan Fang - Binghamton University (United States)
Abstract: The human microbiome plays an important role in human health and disease status. Using next-generation sequencing technologies, it is possible to quantify microbiome composition. Clustering microbiome data can provide valuable information by identifying underlying patterns across samples as well as between the microbes. We develop a novel family of mixtures of logistic normal multinomial models using a modified factor analyzer structure to cluster both the samples and taxa simultaneously. Parameter estimation is done using a variational variant of the alternating conditional expectation conditional maximization(AECM) algorithm that utilizes a variational Gaussian approximation. The proposed method will be illustrated using simulated and real datasets.