Comparisons of computational methods for differential alternative splicing detection using RNA-seq in plant systems

Area of Science:

• Genomics
• Bioinformatics
• Molecular Biology

Background:

• Alternative splicing (AS) is a key post-transcriptional regulation mechanism in eukaryotes.
• Existing AS detection tools are often optimized for animal data, necessitating evaluation for plant-specific gene structures.
• Understanding AS in plants is crucial for crop improvement and fundamental biology.

Purpose of the Study:

• To benchmark popular computational methods for detecting differential alternative splicing (AS) in plants.
• To guide biologists in selecting appropriate tools for plant RNA-seq data analysis.
• To assess the impact of varying data parameters and annotation quality on AS detection.

Main Methods:

• Comparison of eight publicly available software packages for differential splicing analysis.
• Utilized simulated and real Arabidopsis thaliana RNA-seq datasets.
• Evaluated performance based on varying AS ratio, read depth, dispersion, AS types, sample size, and annotation accuracy.
• Validated findings on real data using PCR.

Main Results:

• No single method demonstrated superior performance across all tested conditions.
• Annotation accuracy significantly influenced the choice of the best-performing method.
• DEXSeq performed well with strong AS signals and accurate annotations in simulated data.
• Cufflinks offered a good precision-recall balance, especially with incomplete annotations.
• MATS excelled in analyzing simple AS events in real RNA-seq data.

Conclusions:

• The optimal tool for plant AS analysis depends on specific experimental conditions and annotation quality.
• Accurate gene annotation is critical for reliable differential splicing detection.
• Complex AS events remain challenging for most current computational methods.