Abstract
BACKGROUND: Phylogenetically informed distances are commonly used in the analysis of microbiome data, and analysts have many options to choose from. Although all phylogenetic distances share the goal of incorporating the phylogenetic relationships among the bacteria, they do so in different ways and give different pictures of the relationships between the bacterial communities. RESULTS: We investigate the properties of two classes of phylogenetically informed distances: the Unifrac family, including weighted, unweighted, and generalized Unifrac, and the DPCoA family, which we introduce here. Through several lines of evidence, including a combination of mathematical, data analytic, and computational methods, we show that a major and heretofore unrecognized cleavage in the phylogenetically informed distances is the relative weights placed on the deep and shallow parts of the phylogeny. Specifically, weighted Unifrac and DPCoA place more emphasis on the deep parts of the phylogeny, while unweighted Unifrac places more emphasis on the shallow parts of the phylogeny. Both the Unifrac and the DPCoA families have tunable parameters that can be shown to control how much emphasis the distances place on the deep or shallow parts of the phylogeny. CONCLUSIONS: Our results allow for a more informed choice of distance and give practitioners more insight into the potential differences resulting from different choices of distance.