Assessing the Probability that a Finding Is Genuine for Large-Scale Genetic Association Studies

Area of Science:

• Genetics
• Statistical genetics
• Bioinformatics

Background:

• Genetic association studies generate numerous P-values, complicating the identification of genuine findings.
• High-throughput sequencing increases the number of variants tested, exacerbating the multiple testing problem.
• Current significance thresholds may not effectively filter out false positives in large genetic datasets.

Purpose of the Study:

• To introduce a novel method, Probability that a Finding is Genuine (POFIG), for directly estimating the probability of a spurious finding from a P-value.
• To provide a statistically sound and easily interpretable alternative to existing methods like False Positive Report Probability (FPRP).
• To address limitations of current methods in assessing individual findings and handling large-scale genetic data.

Main Methods:

• Development of the POFIG statistical method to derive the probability of a spurious finding from an observed P-value.
• Comparison of POFIG's properties with existing methods, focusing on interpretability and validity.
• Application of the POFIG method to analyze Genome-Wide Association Study (GWAS) data for Crohn's disease.

Main Results:

• POFIG provides a valid probability that a genetic finding is spurious, directly interpretable from a P-value.
• The method is robust, immune to multiple testing and selection bias, and can be averaged across multiple findings.
• Analysis of GWAS data for Crohn's disease demonstrated the practical application and utility of POFIG.

Conclusions:

• POFIG offers a simple yet statistically rigorous approach to assess the reliability of genetic findings.
• The method enhances the interpretation of P-values in the context of large-scale genetic association studies.
• POFIG facilitates more accurate identification of genuine genetic associations and reduces the impact of false discoveries.