Abstract
Research on the human microbiome has generated a staggering amount of sequence data, revealing variation in microbial diversity at the community, species (or phylotype), and genomic levels. In order to make this complexity more manageable and easier to interpret, new units-the metagenome, core microbiome, and enterotype-have been introduced in the scientific literature. Here, I argue that analytical tools and exploratory statistical methods, coupled with a translational imperative, are the primary drivers of this new ontology. By reducing the dimensionality of variation in the human microbiome, these new units render it more tractable and easier to interpret, and hence serve an important heuristic role. Nonetheless, there are several reasons to be cautious about these new categories prematurely "hardening" into natural units: a lack of constraints on what can be sequenced metagenomically, freedom of choice in taxonomic level in defining a "core microbiome," typological framing of some of the concepts, and possible reification of statistical constructs. Finally, lessons from the Human Genome Project have led to a translational imperative: a drive to derive results from the exploration of microbiome variation that can help to articulate the emerging paradigm of personalized genomic medicine (PGM). There is a tension between the typologizing inherent in much of this research and the personal in PGM.