Controlling the Rate of GWAS False Discoveries

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Controlling the false discovery rate (FDR) is essential for multiple comparison adjustment in genetic association studies.
• Existing FDR methods assume a one-to-one correspondence between unique discoveries and rejected hypotheses, which is often violated in genome-wide association studies (GWAS) due to linkage disequilibrium.

Purpose of the Study:

• To propose a novel approach for FDR control in GWAS that accounts for the aggregation of signals from linked single nucleotide polymorphisms (SNPs).
• To adapt existing FDR-controlling strategies to handle the initial selection of distinct hypotheses based on prescreening.

Main Methods:

• Developed a novel FDR control method incorporating prescreening to define the resolution of distinct hypotheses.
• Adapted FDR-controlling strategies to accommodate this prescreening step.
• Validated the approach using theoretical results and simulations mimicking GWAS dependence structures.
• Tested the method's versatility with single-marker tests and multiple regression analyses.

Main Results:

• The proposed prescreening-based FDR control method effectively addresses the inflation of FDR caused by a posteriori aggregation of signals in GWAS.
• The approach is shown to be versatile and performs well with different analytical methods.
• An R package is provided for practical application on standard GWAS data.

Conclusions:

• The novel FDR control method offers a more accurate way to manage false discoveries in GWAS by accounting for linkage disequilibrium.
• This approach enhances the reliability of genetic association findings and facilitates better understanding of complex traits.