OrthoFinder: improved phylogenetic orthology inference with enhanced accuracy and scalability
TL;DR
OrthoFinder v3 introduces a scalable, two-step phylogenetic algorithm that increases orthogroup inference accuracy by 5-7% and can process over 4,000 genomes, overcoming the quadratic time complexity bottleneck of previous methods.
Problem / question
Identifying evolutionarily related genes across different species traditionally requires all-versus-all sequence comparisons, which scale quadratically and become computationally impossible as global projects sequence thousands to millions of new genomes.
Methods
The authors updated OrthoFinder (v3) with a phylogenetic delineation step that splits erroneously fused orthogroups using gene tree-species tree reconciliation. They also added a linear-scaling workflow using the SHOOT profile algorithm to assign new genomes to a core set of 64 species. They benchmarked v3 against 13 tools (including FastOMA and SonicParanoid2) using OrthoBench, Quest for Orthologs (QfO) datasets, and up to 4,096 Ensembl bacterial genomes.
Key findings
OrthoFinder v3 improved orthogroup inference accuracy (f-score) by 5-7% over version 2. It was the only tool able to process 1,024 eukaryotic genomes within a 7-day limit (completing in 128 hours) and successfully scaled to 4,096 bacterial genomes in 13.6 days using 504 GB of RAM. Compared to FastOMA, OrthoFinder v3 achieved 80% higher recall (15,721 versus 8,686) on the QfO eukaryotic species tree discordance test while maintaining comparable precision.
Why it matters
This update allows researchers to accurately trace gene evolution and transfer biological knowledge across thousands of newly sequenced species using standard computing resources, directly supporting massive biodiversity initiatives like the Earth BioGenome Project.
Limitations
The method is currently capped at analyzing thousands of species, meaning further algorithmic improvements are still needed to reach the ultimate goal of analyzing all millions of species on Earth simultaneously.
Takeaway
By assigning new genomes to a pre-computed core set and using phylogenetic trees to refine gene groupings, OrthoFinder v3 achieves near-linear scalability and top-tier accuracy for massive comparative genomics projects.