An empirical Bayes approach to normalization and differential abundance testing for microbiome data

摘要

It is well known that microbes interact with their human host. The human microbiome, which refers to the collection of microbes and their genetic information in the human body, contributes to healthy human physiology and development, and dysbiosis of microbial communities is linked to many diseases, such as obesity, type 2 diabetes, and inflammatory bowel disease [ – ]. Host genetics and environmental factors, in turn, affect the health and diversity of the human microbiome [ , ]. However, the mechanisms underlying human health and disease remain largely unknown because of the complexity and dynamics of microbial communities. In order to understand the taxonomic composition and biological function of microbiomes, high-throughout sequencing technologies and advanced bioinformatics tools are now routinely employed in microbiome studies [ ]. For example, marker gene analysis involves extracting DNA from primary samples, sequencing a highly variable region, and clustering sequence reads into Operational Taxonomic Units (OTUs) by sequence similarity (e.g., 97%). The evolutionary relationships among OTUs can also be inferred, by using a reference database, or by inferring the phylogenetic tree de novo [ ].