ETE 3: Reconstruction, Analysis, and Visualization of Phylogenomic Data

1The Environment for Tree Exploration (ETE) is a toolkit developed to facilitate the computation, analysis and visualization of phylogenetic data. ETE provides a comprehensive Python programming library (API) that allows researchers to automate common tasks in comparative genomics. Since its first release, ETE has been widely used as a computational framework to perform numerous phylogenomic analyses, including characterizing newly sequenced genomes, extracting information from large sets of phylogenetic trees and developing third party tools and databases. Here, we describe the latest version of the software (ETE v3), featuring a significantly improved API library and a novel collection of standalone tools. While the API continues to offer full programmatic control on data analysis and visualization, the new standalone tools facilitate the use of common phylogenetic methods at the genomic scale. We here describe the most notable additions.

2The ete-build tool provides a unified interface to wrap the execution of reproducible phylogenetic workflows, comprising the reconstruction of gene–trees and supermatrix-based species trees. To do so, ETE relies on a versioned collection of external tools that are transparently installed and executed upon request. A single command is used to configure and launch complex phylogenetic pipelines, covering sequence alignment, trimming, substitution-model testing, tree inference, and image rendering (fig. 1A). In addition, the supermatrix-based reconstruction mode permits to build and concatenate multiple sequence alignments with ease, simplifying the inference of species trees based on multiple genes. Advanced options allow to automatically switch from amino-acid to nucleotide alignments based on sequence identity, resuming the execution of workflows, or even testing multiple strategies in parallel. As an example, a single command line can be used to test several alignment methodologies or phylogenetic inference programs simultaneously, making the tool particularly suitable to run phylogenomic pipelines. Notably, ETE-build was recently used to compute over one million phylogenetic trees for the EggNOG v4.5 database.

3Measuring selective pressures on molecular sequences is a common task in evolutionary biology. Softwares such as CodeML or SLR provide the statistical and computational framework to perform these analyses. However, the use of such tools at the genomic scale requires substantial work on data preparation, on experimental design, and on results interpretation. To aid in these tasks, the ete-evol tool automates CodeML/SLR-based analyses by using pre-configured evolutionary models and directly producing a graphical representation of the results. These pre-configured models include site, branch, branch-site, and clade models. For instance, ete-evol can test, in parallel, and with a single call, the differential selective pressures along each branch in a given phylogeny. Importantly, fitted models are compared using a built-in likelihood ratio test. Evolutionary measures from the best-fitting models are then plotted (or interactively visualized) by mapping the predicted selective pressures acting on sites and branches into the tested topology, as well as on the multiple sequence alignment (fig. 1B). For convenience, raw output files produced by CodeML and SLR can also be visualized using ete-evol.

4ETE v3 provides three measures to compute distances between trees, namely the Robinson–Foulds distance, a branch congruence measure (%) and the TreeKO Speciation distance. In contrast to existing software, ete-compare calculates all three distances at the same time; it accepts trees varying in size and containing duplication events; it allows filtering branches with low support; and it is optimized for comparing large datasets. In addition, ete-compare can provide a detailed list of the differences and coincidences among the compared trees for further analysis. Conveniently, the TreeKO method for splitting gene trees into duplication-free subtrees has been optimized and integrated into ETE’s API library, thereby enabling its use for other tests. For instance, ETE allows summarizing the phylogenetic signal (i.e., gene tree support) from an heterogenous sample of gene trees using a species tree topology as reference (fig. 1C).

5Efficient queries to the NCBI-taxonomy database are now available through the ete-ncbiquery tool or the relevant methods in the API. Extracting pruned subtrees, converting NCBI taxids into their corresponding scientific names, obtaining full lineage tracks, or annotating user-trees with taxonomic data, are common tasks that can be easily performed with the ete-ncbiquery tool. Importantly, all queries are carried out locally, avoiding unnecessary lags and permitting the integration of the tool into genomic and metagenomic pipelines.

6Finally, other ETE-tools and methods are available that aid in routine tasks such as format conversion, topology manipulation, and custom visualization of trees linked to multiple sequence alignments (fig. 1D).

7Although several software packages are available for the standalone exploration of trees and the programmatic manipulation of data, ETE offers a unified framework to compute and analyze genome-wide collections of evolutionary data while providing unique visualization capabilities. Moreover, with the recent addition of the command line tools, ETE has significantly broadened its scope, simplifying many common tasks in phylogenomics for both expert and casual users.