Optimized detection of differential expression in global profiling experiments: case studies in clinical...

Area of Science:

• Biostatistics
• Bioinformatics
• Genomics
• Proteomics

Background:

• Identifying differentially expressed molecular markers is crucial for large-scale profiling studies (e.g., DNA microarrays, mass spectrometry).
• Selecting appropriate statistical methods for diverse high-throughput datasets remains a significant challenge for researchers.
• Existing methods often struggle to balance sensitivity and specificity in marker discovery.

Purpose of the Study:

• To demonstrate the practical application and benefits of the Reproducibility-Optimized Test Statistic (ROTS) in identifying reliable molecular markers.
• To showcase ROTS's effectiveness in both transcriptomic and proteomic data analysis.
• To highlight ROTS's advantages over conventional statistical procedures for differential expression analysis.

Main Methods:

• Utilized ROTS, an adaptive statistical procedure that optimizes test statistics directly from data.
• Applied ROTS to a public leukemia gene expression microarray dataset.
• Applied ROTS to a liquid chromatography-mass spectrometry (LC-MS) dataset of plasma samples from severe burn patients.

Main Results:

• ROTS improved the sensitivity of gene marker lists while maintaining high specificity in the leukemia microarray dataset.
• ROTS identified several peptide markers in the LC-MS proteomic dataset that were missed by conventional analysis.
• Demonstrated ROTS's effectiveness for global quantitative proteomic studies.

Conclusions:

• ROTS offers a robust and adaptive approach for reliable marker discovery in transcriptomic and proteomic studies.
• The procedure enhances the identification of significant molecular markers, improving diagnostic and prognostic potential.
• Freely available implementations of ROTS (R-package, Chipster) promote its widespread adoption in clinical research.