Systematic evaluation of differential splicing tools for RNA-seq studies.

Area of Science:

• Genomics
• Bioinformatics
• Molecular Biology

Background:

• Differential splicing (DS) is a crucial post-transcriptional process impacting cellular functions and diseases.
• Numerous computational tools exist for DS analysis from RNA-seq data, but a comparative evaluation is lacking.

Purpose of the Study:

• To systematically evaluate and compare 10 distinct computational tools for differential splicing analysis.
• Assess tools based on consistency, reproducibility, precision, recall, false discovery rate, and agreement in identifying differentially spliced genes and functional enrichment.

Main Methods:

• Evaluated 10 DS analysis tools representing three categories: exon-based, isoform-based, and event-based methods.
• Selected tools included DEXSeq, edgeR, JunctionSeq, limma, cuffdiff2, DiffSplice, dSpliceType, MAJIQ, rMATS, and SUPPA.
• Performance metrics included consistency, reproducibility, precision, recall, FDR, and gene overlap.

Main Results:

• Exon-based methods and two event-based methods (MAJIQ, rMATS) generally performed well across selected measures.
• Exon-based tools outperformed isoform-based and event-based methods in overall performance.
• Significant variation in tool performance was observed across different datasets and sample sizes.

Conclusions:

• Exon-based computational approaches are generally recommended for differential splicing analysis.
• The choice of DS analysis tool can significantly impact results, highlighting the need for careful selection based on data characteristics.
• Further research is needed to understand the variability in tool performance across diverse biological contexts.