Assessing the Impact of Uncertain Gene Tree Rooting on Phylogenetic Reconciliation Using a Simulation Framework

Evolutionary trees are fundamental to the study of evolution. The two main types of evolutionary trees are gene trees and species trees. A gene tree shows the evolutionary history of a chosen gene family, while a species tree shows the evolutionary history of a chosen collection of species. Duplication-Transfer-Loss (DTL) reconciliation is one of the most effective techniques for studying the evolution of gene families and inferring evolutionary events such as gene duplications, horizontal gene transfers, and gene losses. Given a gene tree and the corresponding species tree, DTL reconciliation compares the gene tree with the species tree and reconciles any differences between the two by proposing gene duplication, horizontal gene transfer, and gene loss events. DTL reconciliations are generally computed using a parsimony framework where each evolutionary event is assigned a cost and the goal is to find a reconciliation with minimum total cost. The resulting optimization problem is called the DTL-reconciliation problem. While standard formulations of the DTL-reconciliation problem require the gene tree and the species tree to be rooted, gene trees can be notoriously difficult to root accurately, and consequently, the gene trees used for DTL reconciliation are often unrooted. When provided with an unrooted gene tree, existing DTL reconciliation algorithms and software first find a root for the unrooted gene tree and then use the resulting rooted gene tree for the reconciliation. The approach employed for rooting unrooted gene trees is to compute the reconciliation cost for each possible rooting of the unrooted gene tree and then choose a rooting that yields the minimum reconciliation cost. However, there is little understanding of the accuracy of DTL reconciliation methods in rooting gene trees. Moreover, gene trees reconciled by DTL reconciliation often have multiple optimal rootings, leading to a great amount of uncertainty regarding their true roots. In the absence of biological data regarding the true root of a gene tree, simulation provides a powerful technique for evaluating the performance of such a rooting method. Furthermore, varying the different parameters, such as the rates of different evolutionary events, used to simulate the evolution of a gene tree can help us identify the conditions under which DTL reconciliation might perform well. Ultimately, the ability to accurately identify the root of a gene tree will allow biologists to construct accurate evolutionary histories and properly understand how genes and species evolve, directly benefiting our understanding of biology. Currently, multiple simulation frameworks with the ability to simulate gene family evolution exist. However, there are several factors that are crucial to the accurate simulation of gene family evolution, especially for prokaryotic gene families, that no existing simulation framework accounts for. In this work, we address this limitation by constructing a simulation framework for gene family evolution that accurately simulates horizontal gene transfers by accounting for both additive and replacing transfers, along with multi-gene transfers, and by improving the implementation of biased gene transfers. We use the new simulation framework to generate a large collection of gene trees and species trees, for different combinations of the parameter settings, and use these trees to test the accuracy of DTL reconciliation in identifying the true root of the gene trees under different evolutionary scenarios.