Geometry of the space of phylogenetic trees with non-identical leaves

Abstract: Phylogenetic trees summarize evolutionary relationships. The Billera-Holmes-Vogtmann (BHV) space for comparing phylogenetic trees has many elegant mathematical properties, but it does not encompass trees with differing leaf sets. To overcome this, we introduce Towering space: a complete metric space that extends BHV space to trees with non-identical leaf sets. Towering space is a structured collection of BHV spaces connected via pruning and regrafting operations. We study the geometry of paths in Towering space and present an algorithm for computing metric distances. By addressing a major limitation of BHV space, Towering space facilitates the analysis of modern phylogenetic datasets such as multi-domain gene trees.