Practical and theoretical considerations in study design for detecting gene-gene interactions using MDR and GMDR...

Area of Science:

• Genetics and Bioinformatics
• Statistical Genomics
• Complex Trait Analysis

Background:

• Identifying interacting genetic risk factors for complex traits presents significant challenges.
• Method selection, sample size, and case-control allocation are critical concerns in study design.
• Existing methods require empirical validation for optimal application.

Purpose of the Study:

• To evaluate the performance of multifactor dimensionality reduction (MDR) and generalized MDR (GMDR) for detecting gene-gene interactions.
• To provide empirical guidelines for study planning and data analysis in gene-gene interaction research.
• To assess statistical power based on accuracy and sample size under various scenarios.

Main Methods:

• Investigated MDR and GMDR performance across diverse experimental scenarios.
• Developed and utilized the mathematical expectation of accuracy as an indicator parameter.
• Examined statistical power within the plausible accuracy range (0.50–0.65) using simulations.

Main Results:

• GMDR with covariate adjustment achieved >80% power in case-control studies with ≥2000 participants and accuracy from 0.56–0.62.
• Sample sizes of ≥4000 were needed for sufficient power when accuracy was <0.56.
• GMDR outperformed MDR for accuracy 0.56–0.62 with sample sizes of 1000–2000.

Conclusions:

• GMDR demonstrates superior performance over MDR when a covariate is adjusted.
• A sample size of 1000–2000 is generally adequate for detecting gene-gene interactions with effect sizes reported in current literature.
• Larger sample sizes are necessary for detecting more subtle genetic interactions with lower accuracy.