SpeciesRax

SpeciesRax infers maximum-likelihood rooted species trees from multiple gene family trees by modeling gene duplication, transfer, and loss and leveraging paralogy and rooting signals in gene trees as phylogenetic information.

Key Features:

• Maximum-likelihood species tree inference: Infers rooted species trees from multiple gene family trees using a maximum likelihood framework.
• Modeling of gene-family events: Explicitly models gene duplication, transfer, and loss, including horizontal gene transfer events, to account for gene/species tree discordance.
• Use of paralogy and rooting signals: Leverages paralogy and rooting signals present in gene trees to improve species tree inference.
• Branch length estimation: Estimates species tree branch lengths in units of expected substitutions per site.
• Branch support via paralogy-aware quartets: Calculates branch support values from paralogy-aware quartets extracted from gene family trees.
• First maximum-likelihood approach for rooted species trees from gene families: Provides a maximum likelihood method capable of inferring a rooted species tree from multiple gene family trees while accounting for duplication, loss, and horizontal gene transfer.
• Performance on large datasets: Demonstrated comparable or superior accuracy and approximately one order of magnitude faster processing on large datasets, exemplified by inferring a rooted vertebrate phylogeny of 188 species from 31,612 gene families in about one hour using 40 computational cores.

Scientific Applications:

• Species tree reconstruction: Reconstruction of rooted species phylogenies from collections of gene family trees.
• Study of gene-family evolution: Analysis of gene duplication, horizontal gene transfer, and gene loss across genomes.
• Resolving gene/species tree discordance: Reconciliation of discordant gene and species trees by accounting for paralogy and transfer.
• Large-scale phylogenomics: Phylogenomic analyses of large datasets, such as vertebrate phylogeny reconstruction from tens of thousands of gene families.

Methodology:

Applies a maximum likelihood framework to infer a rooted species tree from multiple gene family trees while explicitly modeling gene duplication, transfer (horizontal gene transfer), and loss; extracts paralogy-aware quartets from gene family trees to compute branch support and estimates branch lengths in units of expected substitutions per site.