fdrtool: a versatile R package for estimating local and tail area-based false discovery rates

Area of Science:

• Genomics
• Proteomics
• Statistical Bioinformatics

Background:

• High-dimensional data analysis necessitates robust false discovery rate (FDR) methodologies.
• Accurate FDR estimation is crucial for interpreting results in genomic and proteomic studies.

Purpose of the Study:

• To introduce 'fdrtool', an R package designed for comprehensive FDR estimation.
• To provide a unified and flexible framework for FDR analysis across diverse test statistics.

Main Methods:

• The 'fdrtool' package accepts various test statistics (P-values, z-scores, correlations, t-scores).
• It simultaneously estimates local and tail area-based FDR values.
• Empirical null models are fitted to account for deviations from theoretical null distributions.

Main Results:

• 'fdrtool' offers a unified approach to FDR estimation applicable to large-scale multiple testing problems (millions of hypotheses).
• The package provides interpretable graphical outputs for enhanced data visualization.
• It supports a wide range of input data types, increasing its versatility.

Conclusions:

• 'fdrtool' offers a flexible and unified FDR estimation scheme for high-dimensional data.
• Its ability to handle diverse test statistics and empirical null models enhances analytical accuracy.
• The package is a valuable tool for researchers in genomics, proteomics, and statistical bioinformatics.