Identifying main effects and epistatic interactions from large-scale SNP data via adaptive group Lasso

Area of Science:

• Genetics
• Biostatistics
• Computational Biology

Background:

• Genome-wide association studies (GWAS) aim to identify single nucleotide polymorphisms (SNPs) linked to complex diseases.
• SNPs can influence disease risk individually (main effects) or through joint (epistatic) interactions.
• Analyzing high-throughput genetic data is challenging due to the large number of SNPs compared to samples, especially for interaction detection.

Purpose of the Study:

• To propose a novel statistical model for analyzing large-scale genetic association studies.
• To simultaneously identify SNPs and their epistatic interactions associated with complex diseases.
• To enhance the accuracy and reduce false positives in genetic association analyses.

Main Methods:

• Development of an Adaptive Group Lasso (AGL) model incorporating sparsity constraints.
• Introduction of an adaptive reweighting scheme to improve sparsity and reduce false positives.
• Treating SNPs and their interactions as factors for grouped identification, enabling flexible analysis of various disease models.

Main Results:

• The AGL model effectively analyzes SNPs and their interactions simultaneously.
• The adaptive reweighting scheme enhances sparsity, reducing false positive findings.
• The grouped identification approach is flexible for detecting SNP interactions in complex disease models.

Conclusions:

• The proposed AGL method demonstrates advantages in analyzing large-scale genetic association data.
• Validation using simulated and real-world datasets (WTCCC) confirms the method's efficacy.
• Computational intensity for genome-wide interaction detection necessitates combination with filtering methods.