Challenges of Adjusting Single-Nucleotide Polymorphism Effect Sizes for Linkage Disequilibrium

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Genome-wide association studies (GWAS) identify disease-associated SNPs, but small effect sizes necessitate large sample sizes.
• Post-GWAS analyses like polygenic risk scores combine SNP effects to increase power.
• Accounting for linkage disequilibrium (LD) is crucial to avoid over-weighting correlated SNPs.

Purpose of the Study:

• To review methods that adjust SNP effect sizes in summary statistics instead of relying solely on LD pruning.
• To mathematically analyze and compare different LD adjustment approaches.

Main Methods:

• Mathematical analysis of LD adjustment methods.
• Comparison of approaches using Tikhonov regularization as a framework.

Main Results:

• Current Bayesian methods offer implicit regularization parameter choice, enhancing convenience but reducing transparency.
• These methods risk incomplete LD correction, particularly in smaller LD blocks.
• Similarities were observed between LD adjustment processes and Tikhonov-regularized ordinary least squares.

Conclusions:

• No single LD adjustment method is universally best.
• For research prioritizing interpretable LD adjustment, such as identifying genomic causes of disorders, mild regularization of effect sizes may be advantageous.