Comparison of variable and model selection methods for genetic association studies using the GAW15 simulated data.

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Variable and model selection are crucial in genetic association studies.
• Bayesian and non-Bayesian methods offer different approaches to statistical modeling.
• Simulated data provides a controlled environment for method evaluation.

Purpose of the Study:

• To compare and evaluate various variable and model selection methods.
• To assess the performance of Bayesian and non-Bayesian approaches in genetic analysis.
• To identify susceptibility single-nucleotide polymorphisms (SNPs) for rheumatoid arthritis and IgM levels.

Main Methods:

• Utilized three replicates of Genetic Analysis Workshop 15 (GAW15) simulated data.
• Analyzed rheumatoid arthritis (RA) affection status (binary) and IgM levels (continuous).
• Adjusted for covariates including sex, age, and smoking status.

Main Results:

• All evaluated methods effectively detected simulated genetic signals for both phenotypes.
• Successfully identified susceptibility SNPs for RA on chromosomes 6 (HLA region) and 18.
• Identified the susceptibility SNP for IgM on chromosome 11.
• Observed a notable rate of false-positive results across multiple methods.

Conclusions:

• Variable and model selection methods demonstrate comparable performance in detecting true genetic signals.
• Careful interpretation is needed due to the prevalence of false positives.
• The study highlights the strengths and limitations of different statistical approaches in genetic research.