A Continuous Correlated Beta Process Model for Genetic Ancestry in Admixed Populations

Area of Science:

• Population Genetics
• Genomics
• Bioinformatics

Background:

• Admixture creates populations and genomes with genetic ancestry from multiple sources.
• Analyzing genetic ancestry is crucial for trait mapping, speciation studies, and conservation.
• Existing methods for local ancestry inference struggle with isolated admixed populations still segregating for ancestry.

Purpose of the Study:

• To develop and test a novel statistical model for inferring local ancestry frequencies in isolated admixed populations.
• To address the analytical gap in current methods for populations with ongoing or recent admixture.
• To provide a tool for understanding genetic variation in admixed populations.

Main Methods:

• Developed a continuous correlated beta process model.
• Explicitly modeled autocorrelations in population-level ancestry frequencies.
• Utilized discriminant analysis of SNP windows to leverage within-individual ancestry blocks.

Main Results:

• Simulated data analyses demonstrated high accuracy of ancestry frequency estimates with low root-mean-square error.
• The proposed method outperformed traditional hidden Markov models, especially for distant admixture events.
• Analyses of human (Uyghur) and mouse hybrid zone populations revealed significant variation in ancestry frequencies.

Conclusions:

• The developed continuous correlated beta process model effectively fills an analytical gap in local ancestry inference.
• The method is reliable and useful for diverse biological applications, including human disease gene mapping and hybrid zone studies.
• A C++ program (popanc) has been developed and is available online.