Post

Phylogenetic Inference and Visualization: Community Insights

In the rapidly evolving field of viral genomics, selecting the appropriate computational framework for phylogenetic analysis is as critical as the data itself. Following a recent Journal Club presentation, members of the RdRp Summit community engaged in a technical exchange that provided a snapshot of the current software landscape used by experts to reconstruct and visualize evolutionary histories.

This post archives those recommendations, ranging from industry standards to niche libraries for specialized genomic architectures.

1. Phylogenetic Tree Construction

The discussion highlighted a clear distinction between rapid, maximum-likelihood (ML) heuristics for routine surveillance and rigorous Bayesian frameworks for complex evolutionary modeling.

  • Routine & Rapid Inference: IQ-TREE 2 remains the community favorite for its speed and robust model selection. For real-time epidemiology, it is frequently paired with TreeTime (Augur) to generate time-resolved phylogenies.
  • Deep Evolutionary Relationships: When resolving deep nodes, PhyloBayes was noted for its ability to handle complex site-heterogeneous models.
  • Phylodynamics & Reassortment: BEAST 1 and BEAST 2 are the go-to frameworks for phylodynamic inference. Specifically, BEAST 2 was highlighted for its specialized packages like CoalRe, which is essential for analyzing reassortment networks in multi-segmented viruses.
ToolCategoryPrimary Use / StrengthsLink
IQ-TREE 2ML InferenceHigh-performance, automatic model selectionWebsite
PhyMLML InferenceClassical, well-validated ML approachWebsite
FastTreeML InferenceOptimized for extremely large datasetsWebsite
MrBayesBayesianStandard Bayesian MCMC inferenceWebsite
PhyloBayesBayesianSpecialized for deep-level phylogeneticsWebsite
BEAST / BEAST 2BayesianPhylodynamics, molecular dating, reassortmentWebsite
TreeTimePost-processingML-based temporal dating and Augur pipelinesGitHub

2. Visualization and Post-Processing

Visualization remains a matter of both aesthetic preference and functional requirement (e.g., handling non-treelike evolution).

  • Aesthetics: baltic (Python) was highly recommended for producing publication-quality figures, though it carries a steeper learning curve.
  • Interactivity: For exploring Nextstrain-style JSON trees, Auspice is the standard. For visualizing complex reassortment networks, IcyTree remains a unique and vital web-based tool.
  • Programmatic Integration: ggtree (R) and ete3 (Python) continue to be the workhorses for researchers integrating tree plotting into automated pipelines.
ToolTypeKey FeatureLink
balticPythonHighly customizable, publication-ready plotsGitHub
ggtreeRSeamless integration with the R/Tidyverse ecosystemWebsite
FigTreeDesktop GUIClassic, user-friendly tree viewing/editingWebsite
iTOLWeb-basedInteractive, easy annotation of large treesWebsite
AuspiceWeb-basedInteractive viewer for Nextstrain datasetsWebsite
IcyTreeWeb-basedSpecialized for reassortment networksWebsite
ete3PythonComprehensive toolkit for tree manipulationWebsite
phylotree.hyphy.orgWeb-basedLightweight, interactive JS viewerWebsite
phyTreeVizPythonMinimalist plotting libraryGitHub
TreeSwiftProcessingUltra-fast tree traversal and manipulationGitHub
DendroPyProcessingProfessional-grade library for phylogenetic dataWebsite
toytreePythonInteractive plotting in Jupyter notebooksGitHub

Join the Community

This summary represents only a fraction of the daily knowledge exchange within our community. If you are a virologist, bioinformatician, or student working with RNA viruses, consider joining our Slack to participate in these discussions in real-time.

Summary by: Shoichi Sakaguchi
Tags: PhylogeneticsTools