Evaluation of methods for differential expression analysis on multi-group RNA-seq count data.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• RNA sequencing (RNA-seq) is crucial for transcriptome analysis and identifying differentially expressed genes (DEGs).
• Existing evaluations of RNA-seq analysis methods primarily focus on two-group comparisons.
• There is a need for comprehensive evaluations of methods for multi-group RNA-seq data analysis.

Purpose of the Study:

• To compare the performance of 12 different pipelines for detecting differential gene expression (DE) in multi-group RNA-seq count data.
• To evaluate these pipelines using both simulated and real count datasets, with a focus on three-group comparisons (with and without replicates).

Main Methods:

• Comparison of 12 differential expression analysis pipelines from nine R packages.
• Evaluation using simulated and real RNA-seq count data for three-group comparisons.
• Assessment of pipeline performance based on accuracy and consistency of DEG identification.

Main Results:

• Pipelines within the TCC package demonstrated comparable or superior performance across various simulation scenarios.
• The TCC package utilizes a multi-step normalization strategy (DEGES), integrating functions from packages like edgeR and DESeq2.
• Significant variation in DEG counts (18.5–45.7%) was observed among pipelines for the same real dataset, yet expression pattern distributions remained similar.
• Differential expression results correlated with hierarchical clustering dendrograms of raw count data.

Conclusions:

• DEGES-based pipelines in TCC perform effectively for both two-group and three-group RNA-seq comparisons.
• For count data with replicates (especially small sample sizes), the EEE-E pipeline (DEGES with edgeR) is recommended.
• For count data without replicates, the SSS-S pipeline (DEGES with DESeq2) is recommended.