Accurate simulation of large collections of phylogenetic trees

Phylogenetic analyses are growing at a rapid rate, producing increasingly large collections of trees. Scientists rarely share their tree collections, making it difficult for researchers to develop methods that anticipate and respond to this growth of data. While common methods for simulating phylogenetic trees focus on random topologies, the tree collections returned from phylogenetic search are rarely random and contain a high degree of topological similarity. In this paper, we introduce TreeSim, a software package that simulates large tree collections from published consensus trees. TreeSim implements our new simulation algorithm, the combined consensus. Our experimental results indicate that simulating trees based on the combined consensus produces collections whose topological diversity most closely resemble the trees returned from phylogenetic search. We expect that TreeSim will play a critical role in guiding the algorithmic development of new approaches that support the growth of phylogenetic data.