Genome-wide association studies using binned genotypes

Area of Science:

• Genomics
• Statistical Genetics
• Bioinformatics

Background:

• Genome-wide association studies (GWAS) traditionally use linear mixed models (LMM) testing one marker at a time.
• This single-marker approach suffers from over-conservativeness, ignores linkage disequilibrium (LD), and can reduce power due to overfitting.
• Multiple locus models are more appropriate for simultaneously analyzing all genome markers.

Purpose of the Study:

• To develop a more powerful and accurate GWAS method by addressing the limitations of single-marker LMM.
• To introduce a novel bin model that leverages linkage disequilibrium (LD) for dimension reduction.
• To compare the performance of the proposed BIN-Lasso method against existing GWAS approaches.

Main Methods:

• Proposed a novel bin model that groups neighboring markers based on their LD relationships.
• Treated each bin as a synthetic marker for association testing.
• Applied penalized multiple regression, specifically the least absolute shrinkage and selection operator (LASSO), to fit all bins in a single model (BIN-Lasso).

Main Results:

• Simulation experiments demonstrated that BIN-Lasso is more powerful and has a lower Type I error rate than SNP-Lasso and Q+K-LMM.
• Application to a Chinese Simmental beef cattle population for bone weight identified more significant associations than classical LMM.
• The bin model effectively utilizes LD information for dimension reduction.

Conclusions:

• The bin model represents a significant advancement in associative genomics, particularly for big data.
• BIN-Lasso offers improved power and accuracy in GWAS by incorporating LD information.
• This dimension reduction technique provides a powerful new tool for genetic association studies.