Seedlabs
Source research

IQ-TREE 3: phylogenomic inference software using complex evolutionary models

Thomas K. F. Wong, Nhan Ly-Trong, Huaiyan Ren, Piyumal Demotte, Hector Baños, Andrew J. Roger et al. · 2026 · 414 citationsRead the paper

IQ-TREE (https://iqtree.github.io/) is a widely used open-source software tool for efficiently inferring phylogenetic trees under maximum likelihood. Here, we present IQ-TREE version 3, the third major release of the software. IQ-TREE 3 significantly extends version 2 with new features, including mixture models as an alternative to partitioned models, gene and site concordance factors to quantify discordance between genomic regions, integration with phylogenomic divergence time estimation, and a fully featured sequence simulator. The IQ-TREE 3 source code is available at https://github.com/iqtree/iqtree3.

What the full text shows

Deployed in practice
  • AliSim sequence simulator outperforms Seq-Gen, Dawg, and INDELible in runtime and memory usage. (Simulating sequences using advanced mixture and complex models)
  • Automatic determination of the number of classes (k) and substitution models for DNA mixture models via MixtureFinder. (DNA data)
  • Integration of CMAPLE for ultra-fast tree inference of millions of closely related sequences. (Pandemic-scale phylogenetics (e.g., SARS-CoV-2))
  • Implementation of the IQ2MC workflow to speed up precalculation for Bayesian divergence time estimation in MCMCTree. (Large phylogenomic datasets)
  • Introduction of clade-specific protein substitution matrices (Q.bird, Q.insect, Q.mammal, Q.plant, Q.yeast) and a general matrix Q.pfam. (Protein data)
Setup
Open-source software release; tested on diverse genomic datasets including SARS-CoV-2 and various eukaryotic/prokaryotic clades.
How
The software implements maximum likelihood phylogenetic inference, integrating the MixtureFinder algorithm for DNA mixture models and the CMAPLE method for large-scale pathogen trees. It adds a built-in alignment simulator (AliSim) and a protein matrix estimation tool (QMaker).
Limitations
Computational time and memory for mixture models increase linearly with the number of classes (k). · The MAST model requires a user-provided set of input tree topologies.

Summarized by Gemma 4 from the open-access full text · 3 Oct 2026. Read it