STAG: Species Tree Inference from All Genes

Abstract Species tree inference is fundamental to our understanding of the evolution of life on earth. However, species tree inference from molecular sequence data is complicated by gene duplication events that limit the availably of suitable data for phylogenetic reconstruction. Here we propose a novel method for species tree inference called STAG that is specifically designed to leverage data from multi-copy gene families. By application to 12 real species datasets sampled from across the eukaryotic domain we demonstrate that species trees inferred from multi-copy gene families are comparable in accuracy to species trees inferred from single-copy orthologues. We further show that the ability to utilise data from multi-copy gene families increases the amount of data available for species tree inference by an average of 8 fold. We reveal that on real species datasets STAG has higher accuracy than other leading methods for species tree inference; including concatenated alignments of protein sequences, ASTRAL & NJst. Finally we show that STAG is fast, memory efficient and scalable and thus suitable for analysis of large multispecies datasets.