Inferring ancestry with the hierarchical soft clustering approach tangleGen.

Area of Science:

• Population Genetics
• Computational Biology
• Machine Learning

Background:

• Understanding genetic ancestry is crucial for human evolutionary history, personalized medicine, and forensics.
• Current methods like ADMIXTURE infer genetic admixture but lack hierarchical interpretation of complex population structures.

Purpose of the Study:

• To introduce tangleGen, a novel soft clustering tool for population genetics.
• To leverage hierarchical machine learning and graph theory for improved interpretation of ancestral relationships.

Main Methods:

• TangleGen applies the Tangles framework, utilizing graph theoretical concepts.
• It employs a hierarchical clustering approach for population structure analysis.
• The tool identifies single-nucleotide polymorphisms (SNPs) responsible for clustering.

Main Results:

• TangleGen provides a hierarchical perspective on population composition and structure.
• It enhances the interpretability of inferred ancestral relationships.
• The tool successfully demonstrates its capabilities on simulated and real-world data (1000 Genomes Project).

Conclusions:

• TangleGen offers a more interpretable and explainable method for inferring genetic ancestry.
• Its hierarchical framework advances the analysis of complex population structures.
• The identification of causal SNPs adds a new layer of biological insight.