Identification of cancer driver exons with potential therapeutic applications

Alternative splicing sustains cancer progression and can be targeted pharmacologically. Yet, identifying potential cancer driver exons genome-wide has been challenging. Here, we propose a method to pinpoint such exons by regressing gene-level cancer dependencies from knockdown viability screens in cancer cell lines against their corresponding splicing profiles and gene expression. Our statistical models predict the effects of single and multiple splicing perturbations on cell proliferation solely from transcriptomic data, facilitating “in silico RNA isoform screening”. This expedites the identification of potential targets for splicing-based anti-tumoral therapies. We uncovered 1,073 potential exons affecting cell proliferation, mainly from genes not previously linked to cancer, which could have a role in cancer progression alongside genes with driver mutations. Experimental validation of exons prioritized from patient-derived tumors spanning 13 cancer types confirms their impact on cell proliferation, specifically in highly proliferative cancer cell lines. Integrating pharmacological screens with predicted splicing dependencies identifies potential cancer driver exons influencing drug sensitivity as well. Additionally, our models for drug-exon interactions show promise in predicting treatment outcomes from patient tumor transcriptomes, suggesting potential applications in precision oncology. This study presents a novel approach for identifying potential cancer driver exons and their possible therapeutic applications, underscoring the importance of alternative splicing as a therapeutic target in cancer.