Identifying latent genetic interactions in genome-wide association studies using multiple traits.

Area of Science:

• Genetics
• Bioinformatics
• Statistical Genetics

Background:

• Genome-wide association studies (GWAS) often report lower heritability estimates compared to family-based studies, a phenomenon known as 'missing heritability'.
• This discrepancy may stem from unobserved or complex genetic interactions (gene-gene or gene-environment) that are challenging to detect with traditional methods.
• Identifying these interactions is crucial for a comprehensive understanding of complex trait genetics.

Approach:

• Proposed Latent Interaction Testing (LIT), a novel method to detect latent genetic interactions by leveraging pleiotropy across multiple related traits.
• LIT capitalizes on the principle that correlated traits influenced by shared genetic interactions exhibit distinct variance and covariance patterns based on genotype.
• Employs a scalable kernel-based framework to analyze trait variance/covariance patterns in relation to genotype, suitable for large biobank datasets and numerous traits.

Key Points:

• Simulations demonstrated that LIT significantly enhances the power to detect latent genetic interactions compared to univariate, trait-by-trait analyses.
• Applied LIT to four obesity-related traits in the UK Biobank, successfully identifying genetic variants with interactive effects.
• Detected variants were located near known genes associated with obesity, validating the method's effectiveness.

Conclusions:

• Latent Interaction Testing (LIT) offers a powerful approach to uncover complex genetic interactions by utilizing shared information across multiple traits.
• The R package 'lit' provides a computationally scalable implementation for analyzing large-scale genomic and phenotypic data.
• LIT advances the field by improving the detection of genetic interactions contributing to the 'missing heritability' in complex traits.