MAST: Phylogenetic Inference with Mixtures Across Sites and Trees

Abstract

Hundreds or thousands of loci are now routinely used in modern phylogenomic studies. Concatenation approaches to tree inference assume that there is a single topology for the entire dataset, but different loci may have different evolutionary histories due to incomplete lineage sorting, introgression, and/or horizontal gene transfer; even single loci may not be treelike due to recombination. To overcome this shortcoming, we introduce the mixture across sites and trees (MAST) model, which uses a mixture of bifurcating trees to represent multiple histories in a single concatenated alignment. The MAST model allows each tree to have its own topology, branch lengths, substitution model, nucleotide or amino acid frequencies, and model of rate heterogeneity across sites. We implemented the MAST model in a maximum-likelihood framework in the popular phylogenetic software, IQ-TREE. Simulations show that we can accurately recover the true model parameters, including branch lengths and tree weights (i.e. frequencies) for a given set of tree topologies. We also show that we can use standard statistical inference approaches to reject a single-tree model when data are simulated under multiple trees (and vice versa). We applied the MAST model to multiple primate datasets and found that it can recover the signal of incomplete lineage sorting in the Great Apes, as well as the asymmetry in minor trees caused by introgression among several macaque species. When applied to a dataset of four Platyrrhine species for which standard concatenated maximum likelihood and gene tree approaches disagree, we find that MAST gives the highest weight to the tree favored by gene tree approaches. These results suggest that the MAST model is able to analyse a concatenated alignment using maximum likelihood, while avoiding some of the biases that come with assuming there is only a single tree. The MAST model can therefore offer unique biological insights when applied to datasets with multiple evolutionary histories. We discuss how it can be extended in the future.

Evolution-inspired engineering of nonribosomal peptide synthetases

“…This pattern suggests infrequent recombination within the C domain and the central segment of the A domain. In line with prior analyses that describe A subdomain swaps as one driver for NRP diversification ( 17 , 18 , 20 – 22 , 31 ), our analysis suggested that recombination events are more frequently observed in the N and C termini of A domains. On the basis of our phylogenetic HMM analysis (Fig.…”

“…Positive values indicated a stronger fit with the first evolutionary history, and negative values, the second history. We verified the robustness of these assignments using an alternative method implemented in IQ-TREE, which is conceptually related to a phylogenetic HMM but instead computes the likelihood as a mixture model across prespecified trees ( 31 ).…”

“…To obtain a more profound understanding of the potential recombination sites identified in our initial analysis, we carried out a second phylogenetic HMM and multitree model approach ( 26 , 31 ), specifically targeting the T domain along with the A-T linker region (Fig. 1B and figs.…”

Species boundaries to the limit: validating species delimitation methods is critical to avoid taxonomic inflation in the genomic era

“…Both trees were rooted with T. spatalurus . We evaluated both topologies by implementing the recently published mixture across sites and trees (MAST) model (Wong et al, 2022). In short, this program calculates topology weights for a number of given topologies (two in our case) across an alignment, calculates the most supported topology at each site, and returns an overall value for the whole alignment (determined by a topology weight ranging between 0 and 1).…”

Lineage diversity in a widely distributed New World passerine bird, the House Wren

“…To increase computation speed and reduce the redundancy in sampling that did not reflect the topological differences of interest between the SNP and mtDNA topologies, we trimmed the topology to 43 samples reflective of the mtDNA and taxonomic diversity in the complex. Second, we used the mixture across sites and trees (MAST) model (Wong et al 2022) in IQTREE2 v.2.2.2.1 (Minh et al 2020), which allows for a mixture of trees from a single concatenated alignment and estimates the weights of different topologies. Using the MAST model, we calculated the weights or the proportion of sites belonging to each of the alternative relationships of T. aedon (monophyletic or paraphyletic).…”