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

Area of Science:

• Genetics
• Bioinformatics
• Statistical Genomics

Background:

• The "missing" heritability in complex traits suggests unobserved genetic interactions.
• Existing methods struggle to detect interactions involving unspecified or unobserved genetic or environmental factors.
• Pleiotropy, where genes affect multiple traits, offers a potential avenue for discovering these interactions.

Purpose of the Study:

• To propose and evaluate a novel kernel-based method, Latent Interaction Testing (LIT), for screening genetic interactions.
• To leverage pleiotropy from multiple related traits to detect latent genetic interactions without specifying the interacting variable.
• To increase the power for detecting genetic interactions compared to traditional univariate approaches.

Main Methods:

• Development of Latent Interaction Testing (LIT), a kernel-based statistical method.
• Utilizing pleiotropy across multiple related traits to infer latent genetic interactions.
• Validation using simulated data to compare LIT's power against univariate methods.

Main Results:

• Simulated data analysis demonstrated that LIT significantly increases the power to detect latent genetic interactions.
• Application of LIT to UK Biobank data identified variants with interactive effects related to obesity.
• Detected interactive variants were located near known obesity-related genes, supporting LIT's biological relevance.

Conclusions:

• Latent Interaction Testing (LIT) provides a powerful new approach for discovering complex genetic interactions.
• The method effectively utilizes pleiotropy to identify genetic variants whose effects are context-dependent.
• LIT has significant implications for understanding the genetic architecture of complex traits, including obesity.