Main Content

Comprehensive Analysis of Alternative Splicing Across Tumors from 8,705 Patients

Cancer Cell. Volume 34, p1-14, August 2, 2018 10.1016/j.ccell.2018.07.001

We present a comprehensive analysis of alternative splicing across 32 TCGA cancer types of 8,705 patients. We detect alternative splicing events (ASE) and tumor variants by reanalyzing TCGA RNA and whole exome sequencing data. Many tumors have thousands of ASE not detectable in TCGA normal and GTEx samples. Overall, tumors have ≈30% more ASE than normal samples. Association analysis of somatic variants with ASE confirmed known trans-associations with variants in SF3B1 and U2AF1 and identified three additional trans-acting variants (IDH1, TADA1, PPP2R1A). On average, we identified ≈940 novel exon-exon junctions (“neojunctions”) in tumor samples not typically found in GTEx normal samples. From protein mass spectra available for TCGA breast and ovarian tumor samples, we confirmed on average ≈1.7 neojunction- and ≈0.6 SNV-derived peptides per tumor sample that are also predicted to be MHC-I binders (“putative neoantigens”). By considering neojunction- in addition to SNV-derived peptides, the fraction of samples for which at least one putative neoantigen can be identified increases from 30% to 75%. Tumor-specific splicing presents a large new class of splicing-associated potential neoantigens that may affect the immune response and could be exploited in immunotherapy, e.g., in personalized tumor vaccines.

Data in the GDC

Supplemental Data

Instructions for Data Download

Open Access Data

  1. Download the appropriate manifest file from the publication page
  2. Use the manifest file to download data using the GDC Data Transfer Tool (DTT) or the GDC API

Controlled Access Data

  1. Download the appropriate manifest file from the publication page
  2. Download a token from the GDC Data Portal
  3. Use the manifest file and token to download data using the GDC DTT or the GDC API

For assistance, please contact the GDC Help Desk: support@nci-gdc.datacommons.io.