Using biological knowledge to discover higher order interactions in genetic association studies

Area of Science:

• Genetics
• Bioinformatics
• Computational Biology

Background:

• Genome-wide association studies (GWAS) have identified genetic variants for common diseases, but explain little heritability.
• A significant portion of heritability may be due to higher-order interactions between single nucleotide polymorphisms (SNPs) or between SNPs and environmental factors.
• Identifying these complex interactions presents substantial statistical and computational hurdles.

Purpose of the Study:

• To propose a novel, scalable method for detecting higher-order gene-gene and gene-environment interactions.
• To address the challenges of missing heritability by incorporating external biological knowledge into a Bayesian framework.
• To enhance the power and specificity of interaction detection in high-dimensional genetic data.

Main Methods:

• Developed a fully Bayesian analysis framework that integrates external biological knowledge.
• Utilized knowledge-driven priors to focus the search on biologically plausible interaction regions.
• Designed the method to be scalable for high-dimensional search spaces, accommodating interactions of any order.
• Compared the proposed method with conventional maximum likelihood approaches using simulated data.

Main Results:

• Simulated data analyses demonstrated improved power, specificity, and effect estimates compared to standard methods.
• Application to breast cancer GWAS data identified interactions enriched for biological processes like growth, metabolic process, and biological regulation.
• The Bayesian approach effectively navigates high-dimensional search spaces by prioritizing biologically relevant interactions.

Conclusions:

• The novel Bayesian method offers a powerful and efficient approach to uncover complex genetic interactions contributing to disease heritability.
• Incorporating external biological knowledge is crucial for overcoming the statistical and computational challenges in identifying higher-order interactions.
• This method advances the field of genetic association studies by providing a more comprehensive understanding of disease etiology.