SVDquartets

SVDquartets infers phylogenetic relationships among quartets of taxa under the coalescent model by analyzing site-pattern frequency matrices with singular value decomposition to score competing quartet topologies.

Key Features:

• Algebraic Statistical Approach: Applies algebraic statistics by performing singular value decomposition on matrices of site-pattern frequencies corresponding to splits, enabling quartet scoring without Bayesian MCMC.
• Scalability: Designed to scale efficiently to large genomic and whole-genome datasets.
• Quartet-Based Inference: Focuses on quartets of taxa to identify best-supported topologies that can be aggregated into species trees.
• Uncertainty Quantification: Incorporates nonparametric bootstrap to quantify statistical support and uncertainty for inferred quartet relationships.

Scientific Applications:

• Species Tree Inference: Aggregates quartet-based topologies to infer species-level phylogenies for larger taxon samples.
• Genomic Studies: Applied to whole-genome datasets and demonstrated in empirical studies such as analyses of Sistrurus rattlesnakes and soybeans.

Methodology:

Constructs matrices from site-pattern frequencies corresponding to splits, applies singular value decomposition to derive quartet scores, and uses nonparametric bootstrap to assess uncertainty.