PSI-Sigma: a comprehensive splicing-detection method for short-read and long-read RNA-seq analysis

Area of Science:

• Bioinformatics
• Molecular Biology
• Genomics

Background:

• Percent Spliced-In (PSI) values are crucial for quantifying alternative pre-mRNA splicing (AS) events.
• Existing PSI-detection tools have limitations, often restricted to specific AS event types and evaluated using simulated data.

Purpose of the Study:

• To develop and evaluate PSI-Sigma, a novel tool for PSI value detection.
• To benchmark PSI-Sigma against leading tools using actual RNA-seq data.

Main Methods:

• Developed PSI-Sigma with a new PSI index.
• Utilized RNA Sequins (spliced synthetic genes) for benchmarking with non-simulated RNA-seq data.
• Compared PSI-Sigma's performance (precision, recall, false positive rate, correlation) against rMATS, SUPPA2, and Whippet.

Main Results:

• PSI-Sigma demonstrated superior performance compared to rMATS, SUPPA2, and Whippet.
• The tool showed particular strength in analyzing AS events with multiple alternative exons and intron retention.
• Initial evaluation indicated effectiveness in long-read RNA-seq analysis.

Conclusions:

• PSI-Sigma offers a more accurate and robust method for quantifying alternative splicing events.
• The tool's performance on real-world data and complex splicing scenarios makes it valuable for RNA-seq analysis.
• PSI-Sigma is available for broader use in the scientific community.