A metric derived from node labeling schemes can compare tree shapes based on the intuition that trees are similar when they share many subtrees with identical labels. This metric successfully groups trees from the same random evolutionary process together in multidimensional scaling plots while distinguishing between trees from different stochastic processes. [@colijn_metric_2018]
Definitions
Synthesis
Phylogenetic metrics based on node labeling schemes provide a mathematical framework for quantifying distances between tree shapes, exploiting the principle that root node labels uniquely encode binary tree topology through systematic characterization of subtree structures. When applied to trees generated from different stochastic evolutionary models, these metrics successfully group trees by their underlying generative processes, distinguishing birth-death models from Yule and Aldous models more effectively than simple summary statistics like tree imbalance. The approach leverages structural features such as cherry subtree frequencies, which relate directly to overall tree asymmetry and can be linked mechanistically to parameters like the basic reproduction number. While the metric demonstrably separates trees from distinct evolutionary scenarios, the specific topological features that best capture subtle differences between similar generative processes remain an active area of investigation.
Related
- Tree asymmetry relates to frequency of symmetric cherry subtrees
- Phylogenetic metrics distinguish trees from different generative processes
- Root node label uniquely defines binary tree shape