CFE 2020: Start Registration
View Submission - CMStatistics
Title: Phylogenetically informed Bayesian truncated copula graphical models for microbial association networks Authors:  Hee Cheol Chung - Texas A&M University (United States) [presenting]
Irina Gaynanova - Texas A and M University (United States)
Yang Ni - Texas A&M University (United States)
Abstract: Microorganisms present in nature often co-occur, forming communities that have been found to play a critical role in host health. The recent development of high-throughput sequencing technologies provide opportunities for a deeper understanding of microbial communities. However, due to limited sequencing depth, microbiome data have a large excess of technical zeros, which poses a statistical challenge for reverse-engineering microbial association networks. We propose a Bayesian graphical model based on a latent Gaussian copula by modeling technical zeros as censored observations of a latent variable. Microbes' evolutionary information is incorporated as a prior distribution for edge inclusion probabilities using the diffusion process and latent position model. Numerical studies based on simulated examples suggest that the phylogenetic tree prior significantly improves estimation performance. We present an analysis of a quantitative microbiome profiling data set, and compare it to existing methods.