Motivation: Distance methods are well suited for constructing massive phylogenetic trees. However, the computational complexity for Rzhetsky and Nei’s minimum evolution (ME) approach, one of the earliest methods for constructing a phylogenetic tree from a distance matrix, remains open.
Results: We show that Rzhetsky and Nei’s ME problem is NP-complete, and so probably computationally intractable. We do this by linking the ME problem to a graph clustering problem called the quasi-clique decomposition problem, which has recently also been shown to be NP-complete. We also discuss how this link could potentially open up some useful new connections between phylogenetics and graph clustering.
Supplementary information: Supplementary data are available at Bioinformatics online.