Potential synthetic associations created by epistasis.

Area of Science:

• Genetics and Genomics
• Statistical Genetics
• Bioinformatics

Background:

• Genome-Wide Association Studies (GWAS) identify genetic variants associated with traits.
• Synthetic associations, where a non-causal variant tags multiple causal variants, are poorly understood.
• Current GWAS methods may overlook complex genetic architectures due to reliance on single-locus or strong linkage disequilibrium assumptions.

Purpose of the Study:

• To explore the prevalence of synthetic associations in human GWAS.
• To develop a novel machine-learning approach for inferring synthetic associations using genotype data.
• To investigate the genetic basis of synthetic associations, distinguishing between common and rare variant contributions.

Main Methods:

• Developed a machine-learning algorithm utilizing only genotype data.
• Applied the method to analyze human GWAS data.
• Quantified the proportion of potential synthetic associations within the GWAS Catalog.

Main Results:

• Estimated that 3-5% of peaks in the GWAS Catalog could represent synthetic associations.
• Found that synthetic associations are frequently driven by epistatic interactions among common variants.
• Indicated that multiple rare variants acting independently are less likely to cause these synthetic signals.

Conclusions:

• Synthetic associations are a relevant phenomenon in human GWAS, impacting interpretation.
• The findings underscore the importance of employing multi-locus models in genetic association studies.
• Emphasized the need for careful interpretation of GWAS results and robust follow-up analyses, including fine-mapping and trait prediction.