Lars Arvestad: A divide-and-conquer heuristic for phylogeny inference is better than NJ heuristics

Abstract

I will describe and discuss, "dnctree", my divide-and-conquer heuristic for distance-based phylogeny inference. It uses the established heuristic Neighbor Joining as a base case on small problem instances, and divide and conquer steps that are both easy to explain and implement. I will show results from computational experiments on simulated and biological data. In particular, there are now results from comparisons with other well-known heuristics showing that dnctree is highly competitive.