Widespread 3’ UTR splicing regulates expression of oncogene transcripts in sequence-dependent and independent manners

RNA-seq data from 7897 TCGA cancer and normal tissue samples were used to generate a cancer specfic transcript assembly using stringtie.We used a custom algorithm based on comparison the a filtered reference transcriptome to annotate transcripts that contained introns in the 3' UTR. This assembly was then used to quantitate the same samples using Salmon to produce per transcript, per patient expression levels. The fractional expression of each transcript was calculated as transcript expression divided by gene expression. rMATs was used to calculate the change in retained intron usage between normal and cancer samples on a custom event set based on the previously identified 3' UTR introns. The counts of reads compatible and incompatible with splicing for these introns from each sample was also recorded. Finally feature counts was used to count reads mapping to each exon on junction counts of each intron.