Estimating odds ratios in genome scans: an approximate conditional likelihood approach.

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Genome-wide association studies (GWAS) utilize stringent statistical thresholds to control for false positives.
• These thresholds can lead to biased estimation of genetic effect sizes, overestimating their true magnitude.
• Accurate estimation of genetic effects is crucial for understanding disease heritability and identifying causal variants.

Purpose of the Study:

• To develop and validate a method for correcting bias in genetic effect size estimation in case-control association studies.
• To provide a practical approach applicable to existing GWAS data without requiring raw data access.
• To improve the reliability of genetic effect size estimates derived from stringent genome-wide scans.

Main Methods:

• The proposed method utilizes standard genetic effect estimates and their standard errors from statistical software.
• It corrects for bias introduced by significance thresholding in genome-wide scans.
• Extensive simulations were performed across various genetic models, allele frequencies, and effect sizes to evaluate performance.

Main Results:

• The novel method significantly reduces bias and mean squared error compared to naive estimation, particularly for modest genetic effects.
• It provides a principled approach for constructing confidence intervals that account for statistical significance.
• Simulations confirm the method's effectiveness across diverse genetic scenarios.

Conclusions:

• The developed method offers a practical and widely applicable solution for correcting biased genetic effect size estimates in GWAS.
• It enhances the accuracy of genetic effect estimations, especially in the context of stringent significance thresholds.
• This approach is valuable for re-analyzing published GWAS data and improving the interpretation of genetic findings.