Our ability to measure the RNA landscape in health and disease is improving rapidly, powered by long-read, single-cell and spatial transcriptomics. Realising these advances requires new computational methods for discovery and interpretation, since the transcripts that matter most in cancer are often the ones absent from reference catalogues. This talk focuses on the altered transcripts that drive cancer: the novel splice variants, gene fusions, and structural rearrangements hidden within tumour transcriptomes and the methods my lab has built to find them.