[The problem of multiple testing and solutions for genome-wide studies]

Area of Science:

• Genomics
• Statistical Genetics
• Bioinformatics

Context:

• Genome-wide studies involve analyzing a vast number of statistical tests.
• Traditional significance thresholds (e.g., p=0.05) can lead to a high rate of false positives when applied across numerous tests.
• Existing multiple testing correction methods (Bonferroni, step-down, step-up, graphical) aim to control false positives but may introduce limitations.

Purpose:

• To critically evaluate the utility and drawbacks of traditional multiple testing correction methods in the context of genome-wide association studies (GWAS).
• To introduce and advocate for the adoption of the False Discovery Rate (FDR) and q-values as a more appropriate statistical framework for genome-wide data analysis.

Summary:

• Discusses the problem of multiple testing in genome-wide studies, where standard p-values can yield a significant proportion of false positives.
• Reviews common correction methods (Bonferroni, step-down, step-up, graphical) and highlights their potential issues, including increased Type II error rates and difficulties in determining the number of tests.
• Proposes the False Discovery Rate (FDR) and q-values as a superior alternative, where a q-value represents the expected proportion of false positives among significant results, offering a more interpretable measure of significance for large-scale genomic data.

Impact:

• Provides a nuanced perspective on statistical significance testing in genomics, moving beyond routine application of traditional corrections.
• Promotes the use of q-values for more accurate interpretation of results in genome-wide studies, potentially leading to more reliable scientific discoveries.
• Offers a framework to mitigate the inflation of false positives and improve the efficiency of hypothesis testing in large-scale genetic research.