Microbiome studies have been accelerated by the development of high throughput sequencing technologies and become one of the most active research areas in biological and biomedical sciences. My group has been working on selecting interesting microbial features to answer different types of biological questions. In this talk, I will give a brief outline of our projects with the focus on one project about differential abundance analysis, i.e., to identify features whose abundance levels change across conditions. The task is challenging because of high dimensionality and sparsity of microbiome data. In addition, microbiome experiments often employ a small sample size. We propose to use Poisson Hurdle distributions in a hierarchical model framework to borrow information across features. We develop a fully Bayesian approach for differential abundance analysis while controlling false discovery rate. Simulation studies demonstrate that our proposed method outperforms other existing methods in terms of statistical power and false discovery rate control. We also apply our method to two plant root microbiome studies.