Ancestral haplotype-based association mapping with generalized linear mixed models accounting for stratification

Area of Science:

• Genetics
• Bioinformatics
• Statistical Genomics

Background:

• Genome-wide association studies (GWAS) often involve stratified populations.
• Mixed models with kinship matrices can correct for population and family structure.
• Generalized linear mixed models (GLMMs) extend this to various data distributions.

Purpose of the Study:

• To extend mixed model methodology to GLMMs for GWAS in stratified populations.
• To incorporate ancestral haplotype association analysis.
• To develop and evaluate a computationally efficient software tool.

Main Methods:

• Application of GLMMs with kinship matrices to account for genetic relatedness.
• Inference of ancestral haplotypes using hidden Markov models.
• Association testing using both single nucleotide polymorphisms (SNPs) and haplotypes.

Main Results:

• The GLMM approach effectively corrects for stratification in simulated data with population and/or family structure.
• Association analysis using ancestral haplotypes demonstrated higher statistical power compared to SNPs in simulations.
• The developed model, GLASCOW, efficiently analyzed a large dataset (4600 individuals, 500,000 SNPs) in under 3 hours with moderate RAM usage.

Conclusions:

• The developed GLMM framework provides a robust method for GWAS in stratified populations.
• Ancestral haplotype analysis enhances the power of genetic association studies.
• The GLASCOW software offers a practical and efficient solution for large-scale genetic analyses.