Genome-wide iterative fine-mapping for non-Gaussian phenotypes.

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Genome-wide association studies (GWAS) identify genomic regions linked to phenotypes.
• Traditional fine-mapping methods struggle with small effect sizes and multiple comparisons, leading to low recall and high false discovery rates (FDR).

Purpose of the Study:

• To develop a novel method, Genome-wide Iterative fiNe-mApping (GINA-X), to improve genetic fine-mapping for non-Gaussian data.
• Address limitations of existing two-stage fine-mapping approaches in GWAS.

Main Methods:

• GINA-X employs an iterative process combining a screening step and a variable selection step.
• The screening step identifies candidate variants and estimates the proportion of null variants.
• The variable selection step refines the list of variants using the null proportion estimate to control genome-wide multiplicity.

Main Results:

• Simulation studies demonstrate GINA-X outperforms competing methods in reducing FDR and increasing recall.
• Case studies on alcohol use disorder and breast cancer show GINA-X yields more focused candidate causal variant lists.
• GINA-X-identified variants exhibit improved predictive performance.

Conclusions:

• GINA-X offers a robust approach for genetic fine-mapping, enhancing the identification of causal variants from GWAS data.
• The method effectively manages multiplicity and improves both precision and recall in variant discovery.
• GINA-X has practical applications in understanding complex traits and diseases.