JCD-DEA: a joint covariate detection tool for differential expression analysis on tumor expression profiles

Area of Science:

• Bioinformatics
• Genomics
• Computational Biology

Background:

• Differential expression analysis of tumor profiles is crucial for experimental validation.
• Selecting features that effectively discriminate patient groups is a key challenge.
• Current methods often rely on individual variable hypothesis testing, viewing classification as a secondary step.

Purpose of the Study:

• To develop a novel software tool for joint covariate detection in differential expression analysis.
• To integrate hypothesis testing with classification-based testing for robust feature selection.
• To enhance the reliability and interpretability of differential expression analysis in tumor profiling.

Main Methods:

• Development of a joint covariate detection tool based on the A5 feature selection strategy.
• Integration of hypothesis testing with classification-based testing.
• Application of a Gaussian mixture model for automatic feature selection.
• Introduction of a projection heatmap for data visualization.

Main Results:

• The developed software effectively performs joint covariate detection for tumor expression profiles.
• Experiments demonstrate the enhanced reliability of the proposed method over traditional approaches.
• The tool successfully identifies features that are both individually and jointly significant.

Conclusions:

• Joint covariate detection provides a stronger basis for variable selection in differential expression analysis.
• The developed software improves the reliability of differential expression analysis for tumor expression profiles.
• The software is publicly available for researchers to utilize and validate findings.