Assessing the Potential of Gene Tree Parsimony for Microbial Phylogenomics

A key challenge in microbial phylogenomics is that microbial gene families are often affected by extensive horizontal gene transfer (HGT). As a result, most existing methods for microbial phylogenomics can only make use of a small subset of the gene families present in the microbial genomes under consideration, potentially biasing their results and affecting their accuracy. One well-known approach for truly genome-scale phylogenomics is gene tree parsimony (GTP), which takes as input a collection of gene trees and finds a species tree that most parsimoniously reconciles with the input gene trees. While GTP based methods are widely used for phylogenomic studies of non-microbial species, their underlying reconciliation models are not designed to handle HGT and, therefore, they cannot be meaningfully applied to microbes. No GTP based methods have yet been developed for microbial phylogenomics. In this work, we (i) design and implement the first GTP based approach, PhyloGTP, for microbial phylogenomics, (ii) use an extensive simulation study to systematically assess the accuracies of PhyloGTP and two other recently developed methods, SpeciesRax and ASTRAL-Pro-2, under a range of different conditions, and (iii) analyze two real microbial datasets with different characteristics. We find that PhyloGTP and SpeciesRax are more accurate than ASTRAL-Pro-2 across nearly all tested conditions, that PhyloGTP and SpeciesRax have similar accuracies overall, but there are conditions under which PhyloGTP consistently outperforms SpeciesRax, and that both PhyloGTP and SpeciesRax can sometimes yield incorrect, misleading phylogenies on complex real datasets.