Inferring on Joint Associations From Marginal Associations and a Reference Sample.

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Genomewide association studies (GWAS) often report marginal association coefficients.
• Genetic fine-mapping commonly uses these marginal estimates with a reference panel for joint regression analysis.
• The uncertainty introduced by using a reference panel instead of the original data is often overlooked.

Purpose of the Study:

• To present a statistically sound method for inferring joint regression coefficients from marginal regressions using a reference panel.
• To address the issue of inflated false discoveries and lack of replicability caused by ignoring reference panel uncertainty.
• To provide a framework for valid statistical inference in genetic fine-mapping.

Main Methods:

• Derivation of the asymptotic distribution for estimated coefficients in a joint regression model with a reference panel.
• Development of a methodology to account for the uncertainty associated with reference panel data.
• Application of the method to both pre-selected regions and data-driven selected regions.

Main Results:

• Ignoring reference panel uncertainty leads to inflated false discovery rates and reduced replicability in genetic fine-mapping.
• The derived asymptotic distribution enables valid statistical inference for joint regression coefficients.
• Simulations and real data examples demonstrate the effectiveness and utility of the proposed methodology.

Conclusions:

• Accurate inference in joint regression models using reference panels is essential for reliable genetic fine-mapping.
• The proposed method corrects for reference panel uncertainty, improving the validity of association studies.
• This approach enhances the reproducibility and accuracy of findings in genetic association analyses.