jSplice: a high-performance method for accurate prediction of alternative splicing events and its application to large-scale renal cancer transcriptome data

Abstract

Motivation: Alternative splicing represents a prime mechanism of post-transcriptional gene regulation whose misregulation is associated with a broad range of human diseases. Despite the vast availability of transcriptome data from different cell types and diseases, bioinformatics-based surveys of alternative splicing patterns remain a major challenge due to limited availability of analytical tools that combine high accuracy and rapidity.

Results: We describe here a novel junction-centric method, jSplice, that enables de novo extraction of alternative splicing events from RNA-sequencing data with high accuracy, reliability and speed. Application to clear cell renal carcinoma (ccRCC) cell lines and 65 ccRCC patients revealed experimentally validatable alternative splicing changes and signatures able to prognosticate ccRCC outcome. In the aggregate, our results propose jSplice as a key analytic tool for the derivation of cell context-dependent alternative splicing patterns from large-scale RNA-sequencing datasets.

Availability and implementation: jSplice is a standalone Python application freely available at http://www.mhs.biol.ethz.ch/research/krek/jsplice

Contact: wilhelm.krek@biol.ethz.ch

Supplementary information: Supplementary data are available at Bioinformatics online.