An improved method for computing q-values when the distribution of effect sizes is asymmetric

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Gene expression data often exhibits asymmetry in test statistic distributions, indicating skewed effect sizes.
• The Storey's q-value method is a standard for identifying differentially expressed (DE) genes and controlling false discovery rate (FDR), but it relies solely on P-values.
• This asymmetry can impact the accuracy of DE gene identification using P-value-based methods.

Purpose of the Study:

• To develop an enhanced method for identifying differentially expressed genes that improves upon the traditional q-value approach.
• To leverage both P-values and the sign of test statistics for more accurate gene ranking.
• To increase the sensitivity in detecting truly differentially expressed genes while maintaining robust FDR control.

Main Methods:

• A novel method is proposed that modifies the q-value approach by incorporating the sign of test statistics.
• The method's performance was evaluated through simulation studies using independent normal data and real microarray data.
• The proposed method was applied to analyze two distinct microarray datasets from plant experiments (thale cress and maize).

Main Results:

• The proposed method demonstrated superior gene ranking compared to the traditional q-value method in simulation studies.
• It identified a higher number of truly differentially expressed genes while effectively controlling the false discovery rate (FDR).
• Analyses of thale cress and maize microarray datasets showcased the practical utility of the enhanced method.

Conclusions:

• The proposed method offers a significant improvement over existing P-value-only approaches for DE gene identification.
• Incorporating test statistic sign enhances the power to detect differentially expressed genes.
• The method provides a valuable tool for analyzing gene expression data, particularly when asymmetry is present.