Combining dependent P-values with an empirical adaptation of Brown's method

Area of Science:

• Bioinformatics
• Statistical Genetics
• Computational Biology

Background:

• Combining P-values from multiple statistical tests is crucial in bioinformatics.
• Handling dependent P-values presents a significant challenge in statistical analysis.
• High-throughput biology generates large, correlated datasets requiring robust P-value combination methods.

Purpose of the Study:

• To introduce and evaluate an empirical adaptation of Brown's method for combining dependent P-values.
• To provide a method suitable for the complex data structures encountered in high-throughput biology.
• To offer a practical and accessible solution for P-value combination in bioinformatics.

Main Methods:

• An empirical adaptation of Brown's method (Empirical Brown's Method - EBM) was developed.
• The method was tested using simulated noisy data.
• Performance was validated using gene expression data from The Cancer Genome Atlas.

Main Results:

• Empirical Brown's Method (EBM) demonstrated superior performance compared to Fisher's method.
• EBM also outperformed other existing approaches for combining dependent P-values.
• The method proved effective on both simulated and real-world biological data.

Conclusions:

• Empirical Brown's Method (EBM) is a powerful and reliable approach for combining dependent P-values.
• EBM offers an improvement over traditional methods, particularly for large-scale biological data.
• The method is implemented in user-friendly formats (Python, R, MATLAB) for broad accessibility.