ASAFE

ASAFE estimates ancestry-specific allele frequencies from bi-allelic SNP genotypes in admixed populations using an Expectation-Maximization (EM) algorithm to account for unphased genotypes and local ancestry.

Key Features:

• Admixed population analysis: Targets admixed populations (e.g., Hispanic Americans) and supports scenarios with three-way admixture.
• EM algorithm implementation: Implements an Expectation-Maximization (EM) algorithm that iteratively estimates ancestry-specific allele frequency distributions.
• Bi-allelic SNP genotype handling: Processes bi-allelic single nucleotide polymorphism (SNP) genotypes as input.
• Unphased genotype resolution: Handles unphased genotypes by estimating ancestry-specific allele frequencies without requiring prior phasing.
• Research applications: Provides ancestry-specific frequency estimates to support more accurate GWAS design and interpretation in admixed cohorts.
• Validation and performance: Validated on simulated datasets, demonstrating low error rates in allele frequency estimation.

Scientific Applications:

• GWAS in admixed populations: Improves design and interpretation of genome-wide association studies (GWAS) by supplying ancestry-specific allele frequencies.
• Association detection power: Increases power to detect genotype–phenotype associations by accounting for ancestry-specific allele distributions.
• Population stratification control: Reduces confounding due to population stratification in association analyses.
• Precision medicine: Informs analyses of disease risk and drug response by characterizing genetic diversity across ancestral backgrounds.

Methodology:

Input data comprise bi-allelic SNP genotypes from admixed individuals and their corresponding local ancestry pairs (ancestries encoded among three possible values); an Expectation-Maximization (EM) algorithm performs iterative E steps to estimate ancestry-specific genotype distributions and M steps to update allele frequency estimates; output is estimated ancestry-specific allele frequencies for each analyzed SNP.