Forward LASSO analysis for high-order interactions in genome-wide association study

Area of Science:

• Genetics
• Bioinformatics
• Animal Breeding

Background:

• Genome-wide association studies (GWAS) traditionally focus on low-order single-nucleotide polymorphism (SNP) interactions with main effects.
• The impact of high-order SNP interactions, particularly those involving SNPs without significant main effects, on quantitative traits remains largely unexplored.
• Standard LASSO methods struggle with the computational demands of analyzing high-order interactions among numerous SNPs.

Purpose of the Study:

• To develop a computational method for identifying high-order SNP interactions that influence quantitative traits.
• To address the limitations of existing methods in analyzing complex genetic architectures, including interactions among SNPs lacking main effects.
• To propose an effective strategy for mapping complex genetic variations in livestock populations.

Main Methods:

• Proposed a novel Forward LASSO analysis method to systematically identify and analyze high-order SNP interactions.
• Utilized linear and generalized linear models within the LASSO framework to handle large-scale SNP data.
• Developed a stage-by-stage approach to shrink non-significant genetic effects to zero, improving computational efficiency.

Main Results:

• Forward LASSO effectively identifies high-order interactions among SNPs, including those without main effects.
• Simulations confirmed the proposed method's efficacy compared to standard LASSO for full models in detecting high-order interactions.
• The method was successfully applied to GWAS data for carcass and meat quality traits in beef cattle.

Conclusions:

• Forward LASSO provides a viable and computationally efficient alternative for GWAS, particularly for mapping high-order genetic interactions.
• This approach enhances the ability to uncover complex genetic architectures underlying quantitative traits in animal populations.
• The study demonstrates the utility of Forward LASSO in identifying genetic factors for economically important traits in beef cattle.