Synthesizing Species Trees from Unrooted Gene Trees

Abstract

Synthesizing species trees from a collection of smaller gene trees is a widely used approach for inferring credible species tree estimates. While corresponding computational problems are typically NP-hard, several of these problems have been effectively addressed by using the parameterized Strict Consensus Approach. This approach is limited to gene trees that are rooted. In practice, however, most gene trees are unrooted, and it is often difficult, if not impossible, to identify accurate rootings. Here, we address this stringent limitation by proposing efficient algorithms that adopt the parameterized Strict Consensus Approach to handle unrooted gene trees. Finally, we demonstrate the performance of our algorithms in a comparative study using empirical and simulated data sets.