Prediction and Quantification of Splice Events from RNA-Seq Data

Area of Science:

• Genomics
• Molecular Biology
• Bioinformatics

Background:

• Analyzing splice variants from short read RNA sequencing (RNA-seq) data is complex.
• Existing methods struggle with unannotated and intricate splice events.

Purpose of the Study:

• To develop and validate a novel genome-guided method for predicting and quantifying splice events from RNA-seq data.
• To enable the analysis of unannotated and complex splice variants.

Main Methods:

• A genome-wide splice graph is constructed from mapped RNA-seq reads.
• Splice events are identified recursively and quantified locally.
• The method's accuracy is assessed using simulated and real RNA-seq data, comparing different read aligners.

Main Results:

• The novel method accurately predicts and quantifies splice variants, including unannotated internal exons.
• Validation in human tissues identified 249 novel internal exons, with high confirmation rates via RT-PCR and RNA-seq.
• Comparison with existing methods (MISO, Cufflinks) shows robust quantification.

Conclusions:

• De novo prediction of splice variants remains valuable, even in well-annotated genomes.
• The developed method enhances the analysis of complex transcriptomes.
• An R/Bioconductor package implementation is freely available.