Statistical test to compare the linkage model and the admixture model based on central limit results

Area of Science:

• Population Genetics
• Statistical Genetics
• Computational Biology

Background:

• The Admixture Model estimates ancestry based on allele frequencies and genomic proportions from ancestral populations.
• It assumes marker independence, which may not hold true for linked loci.
• The Linkage Model incorporates linkage disequilibrium between neighboring loci.

Purpose of the Study:

• To extend the Admixture Model by developing a Linkage Model that accounts for genetic linkage.
• To establish theoretical guarantees for statistical inference within the Linkage Model.
• To develop and validate a statistical test for model selection between Admixture and Linkage Models.

Main Methods:

• Developed a Hidden Markov Model (Linkage Model) extending the Admixture Model.
• Proved consistency and asymptotic normality of maximum likelihood estimators for individual ancestry.
• Derived an asymptotic level-α-test for model selection between the Admixture and Linkage Models.

Main Results:

• Established theoretical properties of estimators in the Linkage Model.
• Demonstrated that the developed statistical test is an asymptotic level-α-test.
• Validated the practical utility of the model selection test using The 1000 Genomes Project data.

Conclusions:

• The Linkage Model provides a more refined approach to ancestry inference by considering linkage.
• The developed statistical test reliably distinguishes between the Admixture and Linkage Models.
• The findings have practical implications for population genetic studies and genomic ancestry analysis.