Searching Genome-Wide Multi-Locus Associations for Multiple Diseases Based on Bayesian Inference

Area of Science:

• Genetics
• Bioinformatics
• Computational Biology

Background:

• Genome-wide association studies (GWASs) utilize single nucleotide polymorphism (SNP) genotyping to explore genotype-phenotype relationships.
• Existing multi-locus methods struggle to detect diverse genetic interactions in complex diseases.
• High-order epistasis detection (≥ 2 SNPs) presents significant computational and analytical hurdles due to exponential complexity.

Purpose of the Study:

• To develop a simple, fast, and powerful method for detecting genome-wide multi-locus epistatic interactions across multiple diseases.
• To address the computational and analytical limitations of current methods for high-order epistasis analysis in GWASs.
• To identify novel genetic interactions relevant to complex diseases using a Bayesian inference approach.

Main Methods:

• Development of a novel Bayesian inference method named DAM (Detecting Associated Multi-locus interactions).
• DAM is designed for efficient and powerful detection of genome-wide multi-locus epistatic interactions.
• Validation using simulated data and application to real-world GWAS datasets (Rheumatoid Arthritis, Type 1 Diabetes).

Main Results:

• DAM demonstrated high power and efficiency in detecting epistatic interactions on simulated data.
• Application of DAM to WTCCC GWAS datasets for Rheumatoid Arthritis and Type 1 Diabetes yielded novel findings.
• The method effectively handles the computational complexity associated with analyzing high-order SNP combinations.

Conclusions:

• DAM provides a suitable and efficient solution for analyzing multi-disease-related genetic interactions in GWASs.
• The Bayesian approach effectively overcomes computational challenges in detecting high-order epistasis.
• The identified novel findings suggest the utility of DAM in uncovering complex genetic architectures of diseases.