TreeMix

TreeMix infers patterns of population splits and mixtures from genome-wide allele frequency data to reconstruct ancestral population graphs representing evolutionary history.

Key Features:

• Statistical Modeling: TreeMix employs a statistical framework using a Gaussian approximation to model genetic drift from genome-wide allele frequency data.
• Graph-Based Inference: The software constructs an ancestral population graph in which nodes represent ancestral populations and edges denote splits or mixtures.
• Application to Diverse Species: TreeMix has been applied to datasets such as 55 human populations and 82 dog breeds and wild canids to reveal complex relationships beyond simple trees.
• Detection of Migration Events: TreeMix identifies migration events, for example inferring that Cambodians have approximately 16% ancestry from a population ancestral to other East Asian groups and detecting wolf ancestry in breeds like the boxer and basenji.
• Admixture Insights: The method provides admixture inferences such as mixed ancestry in East Asian toy dog breeds (Shih Tzu and Pekingese) resulting from mixing between modern toy breeds and ancient Asian breeds.

Scientific Applications:

• Population genetics: Inferring historical splits, mixtures, and ancestry proportions among populations using allele frequency data.
• Evolutionary biology: Identifying migration events and admixture processes that shape genetic diversity across species.
• Phylogenetics: Extending bifurcating tree models by reconstructing ancestral population graphs that incorporate gene flow.

Methodology:

TreeMix uses genome-wide allele frequency data within a statistical framework that applies a Gaussian approximation to model genetic drift and infers an ancestral population graph with nodes as ancestral populations and edges representing splits or mixtures.