Transcriptome-wide identification and study of cancer-specific splicing events across multiple tumors
Metrics: PDF 3320 views | HTML 2695 views | ?
Yihsuan S. Tsai1, Daniel Dominguez2, Shawn M. Gomez1,2,3,4, Zefeng Wang1,2
1Curriculum in Bioinformatics and Computational Biology, University of North Carolina, Chapel Hill, NC 27599, USA
2Department of Pharmacology, University of North Carolina, Chapel Hill, NC 27599, USA
3Computer Science, University of North Carolina, Chapel Hill, NC 27599, USA
4Joint Department of Biomedical Engineering at UNC-Chapel Hill and NC State University, Chapel Hill, NC 27599, USA
Shawn M. Gomez, e-mail: [email protected]
Zefeng Wang, e-mail: [email protected]
Keywords: Alternative splicing, Cancer classification, RNA processing, Cell cycle
Received: December 02, 2014 Accepted: January 12, 2015 Published: February 05, 2015
Dysregulation of alternative splicing (AS) is one of the molecular hallmarks of cancer, with splicing alteration of numerous genes in cancer patients. However, studying splicing mis-regulation in cancer is complicated by the large noise generated from tissue-specific splicing. To obtain a global picture of cancer-specific splicing, we analyzed transcriptome sequencing data from 1149 patients in The Cancer Genome Atlas project, producing a core set of AS events significantly altered across multiple cancer types. These cancer-specific AS events are highly conserved, are more likely to maintain protein reading frame, and mainly function in cell cycle, cell adhesion/migration, and insulin signaling pathways. Furthermore, these events can serve as new molecular biomarkers to distinguish cancer from normal tissues, to separate cancer subtypes, and to predict patient survival. We also found that most genes whose expression is closely associated with cancer-specific splicing are key regulators of the cell cycle. This study uncovers a common set of cancer-specific AS events altered across multiple cancers, providing mechanistic insight into how splicing is mis-regulated in cancers.
All site content, except where otherwise noted, is licensed under a Creative Commons Attribution 3.0 License.