- Buchanan 1930
Title: A Bayesian Sparse Multivariate Regression Model with Asymmetric Nonlocal Priors for Microbiome Data
We propose a Bayesian sparse multivariate regression method to model relationship between abundance of microbes and environmental factors for microbiome data. We model abundance counts of a microbe with a negative binomial distribution and relate covariates to the counts through regression. Extending the conventional nonlocal priors, we construct asymmetric nonlocal priors for regression coefficients to efficiently identify relevant covariates with their effect direction. We build a hierarchical model to facilitate pooling of information across microbes and achieve parsimonious models with improved accuracy. We present simulation studies that compare performance of variable selection under the proposed model to those under Bayesian sparse regression models with asymmetric and symmetric local priors and two frequentist models. The simulations show that the proposed model does a good job of identifying important covariates and yields coefficient estimates with favorable accuracy compared with the alternatives. The proposed model is applied to analyze an ocean microbime data collected over time to study association of harmful algal bloom with microbial communities.
Jointly with Kurtis Shuler (Statistics) and Marilou Sison-Mangus (Ocean Sciences).
I have been working as an assistant professor in the Department of Statistics at University of California Santa Cruz (UCSC). I completed my Ph.D. in statistics at The Ohio State University (OSU) in June 2010 and then worked as a postdoctoral fellow in the Department of Biostatistics at The University of Texas M.D. Anderson Cancer Center (UT MDACC) for two years from 2010 to 2012 and as a visiting assistant professor in the Department of Statistics at OSU for the following year. My research interest lies in the development and application of (parametric/nonparametric) Bayesian models. My development of novel methodology is motivated by important problems in (medical) biology and clinical trial design, and by more routine questions that arise in data analysis. The new methodology has been used in high profile settings.