What Are Microbiome Data?

Abstract

In this chapter, we first introduce microbiome data from sources of sequencing in Sect. 2.1. Then, we describle microbiome data structure and provide several real data tables to illustrate the data structure in Sect. 2.2. The features of microbiome data are summarized in Sect. 2.3. Section 2.4 provides a real example to highlight over-dispersed and zero-inflated features of microbiome data. We describle some challenges of modeling microbiome data in Sect. 2.5. Section 2.6 is summary. 2.1 Microbiome Data Microbiome data is generated through 16S rRNA gene sequencing and shotgun metagenomic sequencing. The bioinformatics tools include the pipeline QIIME and mothur. For example, after preprocessing the raw sequences, two ways are available to generate analyzable microbiome data (Chen et al. 2013; Li 2015). The 16S sequences are either mapped to an existing phylogenetic tree in a taxonomy-dependent way (Matsen et al. 2010) or clustered into OTUs (operational taxonomic units) according to similarity in a taxonomic-independent way (Schloss et al. 2009; Caporaso et al. 2010). The first way uses the existing phylogenetic tree structure to generate microbiome datasets, whereas the second way clusters sequence reads based on similarity level and then assigns them to different taxonomic levels. In the second way, the reads from the amplicons are clustered into OTUs, based on sequence similarity, and then OTUs are hierarchically assigned to a taxonomic tree at the kingdom, phylum, class, order, family, genus and species ranks (Liu et al. 2008; Shi and Li 2017) using available methods for accurate taxonomy assignments , including BLAST (Altschul et al. 1990), the online Greengenes (DeSantis et al. 2006) and RDP (Cole et al. 2003) classifiers, and phylogenetic tree-based and multimer clustering tree-based methods. Liu et al. compared these methods and recommended use of Greengenes or RDP classifier (Liu et al. 2008).