LINADMIX

LINADMIX estimates admixture proportions of modern populations from ancient genotypes by solving a constrained linear model in a reduced-dimensional genotype space.

Key Features:

• Constrained Linear Model Solution: Solves a constrained linear model representing ancient and modern genotypes to estimate admixture relationships.
• Representation in Reduced-Dimensional Space: Represents both ancient and modern genotypes in a low-dimensional genotype space to simplify relationships prior to model fitting.
• Estimation of Mixing Coefficients and Standard Errors: Produces estimates of mixing coefficients that quantify contributions of ancient populations and reports their standard errors.
• Model Fit Evaluation (P-value): Computes a P-value to assess how well a proposed admixture model fits the observed genetic data.
• Robustness via Simulation: Performance has been evaluated through extensive simulated studies demonstrating robustness to variation in population size, genetic drift, proportion of missing data, and model misspecification.

Scientific Applications:

• Admixture inference: Estimating admixture proportions of modern populations as mixtures of ancient populations using genotype data.
• Population history reconstruction: Reconstructing population histories and tracing gene flow events between ancient and modern groups.
• Model validation: Assessing statistical support for proposed admixture models via mixing coefficient estimates, standard errors, and P-values.
• Anthropology and population genetics studies: Investigating genetic continuity and divergence over time in bioinformatics, genetics, and anthropology contexts.

Methodology:

Represents ancient and modern genotypes in a reduced-dimensional space and solves a constrained linear model to estimate mixing coefficients and their standard errors, computes a P-value for model fit, and evaluates performance using extensive simulated studies testing robustness to population size, genetic drift, missing data, and model misspecification.