Prog-Plot - a visual method to determine functional relationships for false discovery rate regression methods.

Area of Science:

• Bioinformatics
• Statistical Genetics
• Computational Biology

Background:

• Multiple test corrections are crucial in differential gene expression analysis to manage the false discovery rate (FDR).
• Current P-value correction methods increasingly use regression models incorporating covariates to enhance test power.
• Specifying these regression models often relies on subjective prior knowledge, lacking objective validation.

Purpose of the Study:

• To present Progressive proportions plot (Prog-Plot), a novel visual tool for identifying the relationship between covariates and P-value proportions.
• To provide an objective method for specifying regression models used in P-value correction for FDR control.
• To move beyond reliance on prior knowledge for regression model selection in FDR analysis.

Main Methods:

• Development of the Progressive proportions plot (Prog-Plot) visualization technique.
• Application of Prog-Plot to assess the relationship between covariates and the proportion of P-values under the null hypothesis.
• Utilizing Prog-Plot to inform the objective specification of regression models for P-value correction.

Main Results:

• Prog-Plot effectively visualizes the functional relationship between covariates and the proportion of P-values consistent with the null hypothesis.
• The tool enables objective assessment of covariate relevance for P-value correction models.
• Demonstrates a data-driven approach to regression model selection, improving upon subjective methods.

Conclusions:

• Prog-Plot offers a valuable, objective visual tool for enhancing P-value correction methods in differential gene expression studies.
• This approach facilitates more accurate FDR control by enabling informed regression model specification.
• The method reduces reliance on prior assumptions, promoting reproducible and robust statistical analysis.