FastSpeciesTree: Fast and Scalable Species tree Inference

Abstract

The generation of species trees from genomic data is an important step in understanding evolutionary relationships across the tree of life. With the increasing availability of genomic data, species tree inference methods which can easily and rapidly produce accurate species trees from large datasets are required to enable downstream analyses of these resources. Here we present FastSpeciesTree, a fully automated species tree inference method designed to leverage advances in rapid pairwise sequence alignment and scalable phylogenetic inference methods. We show that FastSpeciesTree is able to produce species trees with accuracies equivalent to trees produced through manual-curation and best-practice approaches. We further demonstrate that FastSpeciesTree can produce these high accuracy phylogenies substantially faster than any competitor method. Finally, we show that FastSpeciesTree is capable of building a species tree comprising almost the entirety of the RefSeq databases available proteomes (n = 33285) in less than 12 hours using standard computing resources. FastSpeciesTree is executed as a single command and can be found at https://github.com/OrthoFinder/FastSpeciesTree.