DeClUt: Decluttering differentially expressed genes through clustering of their expression profiles

Area of Science:

• Transcriptomics
• Bioinformatics
• Computational Biology

Background:

• Differential expression analysis is crucial for identifying biomarkers in transcriptomic studies using next-generation sequencing.
• Existing methods like DESeq2 and edgeR have limitations, including high False Discovery Rates (FDR) and lack of consensus on optimal performance across different datasets.
• Identifying reliable differentially expressed genes (DEGs) is essential for disease diagnostics and prognostics.

Purpose of the Study:

• Introduce DeClUt, a novel algorithm for differential gene expression analysis.
• Address the limitations of existing DEG identification methods, particularly high FDR and inconsistent results.
• Enhance the accuracy and reliability of DEG analysis in transcriptomic studies.

Main Methods:

• DeClUt employs a clustering-based approach, assuming DEGs form compact, well-separated clusters.
• The algorithm's clustering method is designed to be robust to outliers common in RNA-seq data.
• Utilizes the average silhouette function to ensure accurate sample-to-condition membership assignment.

Main Results:

• DeClUt demonstrated superior self-consistency and a significantly lower False Positive Rate (FPR) compared to established methods (edgeR, DESeq2, NOISeq, SAMseq).
• Benchmarked on real RNA-seq datasets, DeClUt outperformed competitors in identifying DEGs.
• In a prostate cancer dataset, DeClUt identified 8 novel DE genes linked to neoplastic processes, missed by other tools.

Conclusions:

• DeClUt offers a novel, clustering-based method for differential expression analysis with improved consistency and reduced FDR.
• The algorithm successfully identifies biologically relevant DEGs missed by current state-of-the-art tools.
• DeClUt has the potential to enhance the efficacy of differential expression analyses across various biological and medical applications.