A model-free approach for detecting interactions in genetic association studies

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Genome-wide association studies (GWAS) have identified single-nucleotide polymorphisms (SNPs) associated with complex traits and diseases.
• Traditional methods often focus on additive genetic models, overlooking genome-wide SNP-SNP interactions due to the vast number of SNPs.

Purpose of the Study:

• To develop an efficient, model-free statistical procedure for detecting SNPs with main genetic effects and epistatic interactions.
• To identify SNPs involved in complex genetic regulatory networks that may be missed by conventional approaches.

Main Methods:

• A two-stage non-parametric independence screening procedure is proposed to identify important main genetic effects and interactions.
• The association between phenotype and genotype is modeled using nonparametric techniques, assuming an unknown function.
• Penalized regressions, such as LASSO, are used to analyze the subset of genetic predictors identified by the screening procedure.

Main Results:

• Simulation studies demonstrate the computational efficiency and strong finite sample performance of the proposed procedure.
• The method successfully selects potential SNPs and identifies significant SNP-SNP interactions.
• A real data analysis highlights the crucial role of epistatic interactions in explaining body mass index.

Conclusions:

• The proposed framework effectively detects SNPs with main effects and epistatic interactions without assuming a specific genetic model.
• This approach is expected to uncover SNPs within genetic regulatory networks previously overlooked.
• The findings underscore the importance of considering epistatic interactions in genetic association studies.