Efficient approaches for large-scale GWAS with genotype uncertainty.

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Cost-efficient genetic association studies are crucial for large-scale research.
• Imputed and low-depth sequencing data present unique challenges for association analysis.

Purpose of the Study:

• To develop and evaluate a novel method for genetic association studies using genotype probabilities.
• To compare the performance of different priors for estimating genotype probabilities.
• To assess the statistical power and bias of the proposed method against existing approaches.

Main Methods:

• Developed ANGSD-asso, a latent variable model within a generalized linear model framework.
• Estimated genotype probabilities using sample allele frequency or individual allele frequencies as priors.
• Compared ANGSD-asso with genotype dosage-based methods using simulations and real data (UK Biobank).

Main Results:

• ANGSD-asso demonstrated higher statistical power and reduced bias compared to genotype dosage methods, particularly in structured populations and with sequencing depth.
• Using individual allele frequencies as a prior improved power over sample allele frequency prior.
• The latent model accommodating additional covariates showed superior performance and speed.

Conclusions:

• ANGSD-asso provides a powerful and efficient approach for genetic association studies with imputed or low-depth sequencing data.
• The latent model framework effectively handles genotype uncertainty and improves statistical power.
• This method offers a significant advancement for large-scale genetic research.