The Cluster Affinity Distance for Phylogenies

Abstract

Studying phylogenetic trees is fundamental to biology and benefitting a vast variety of other research areas. Comparing such trees is essential to such studies for which a growing and diverse collection of tree distances are available. In practice, tree distances suffer from problems that can severely limit their applicability. Notably, these distances include the cluster matching distance that is adapted from the Robinson-Foulds distance to overcome many of the drawbacks of this traditional measure. However, at the same time, the cluster matching distance is much more confined in its application than the Robinson-Foulds distance and makes sacrifices for satisfying the properties of a metric. Here, we propose the cluster affinity distance, a new tree distance that is adapted from the cluster matching distance but has not its drawbacks. Nevertheless, as we show, the cluster affinity distance preserves all of the properties that make the matching distance appealing.