An algorithm for reconstructing ultrametric tree-child networks from inter-taxa distances.

Abstract

Traditional “distance based methods” reconstruct a phylogenetic tree from a matrix of pair-wise distances between taxa. A phylogenetic network is a generalisation of a phylogenetic tree that can describe evolutionary events such as reticulation and hybridisation that are not tree-like. Although evolution has been known to be more accurately modelled by a network than a tree for some time, only recently have efforts been made to directly reconstruct a phylogenetic network from sequence data, as opposed to reconstructing several trees first and then trying to combine them into a single coherent network. In this work we present a generalisation of the UPGMA algorithm for ultrametric tree reconstruction which can accurately reconstruct ultrametric tree-child networks from the set of distinct distances between each pair of taxa.