Abstract
The interactions among non-coding RNA (ncRNA) and RNA binding protein (RBP) are increasingly recognized as one of basic mechanisms in gene regulation, and play a crucial role in cancer progressions. However, the current understanding of this regulation network, especially its dynamic spectrum according to the differentially expressed nodes (i.e. ncRNAs and RBP) is limited. Utilizing transcriptomics and interactomics resources, dysregulated RBP-ncRNA circuits (RNCs) are systematically dissected across 14 tumor types. We found these aberrant RNCs are robust and enriched with cancer-associated ncRNAs, RBPs and drug targets. Notably, the nodes in altered RNCs can jointly predict the clinical outcome while the individual node can't, underscoring RNCs can serve as prognostic biomarkers. We identified 30 pan-cancer RNCs dysregulated at least in six tumor types. Pan-cancer RNC analysis can reveal novel mechanism of action (MOA) and repurpose for existing drugs. Importantly, our experiments elucidated the novel role of hsa-miR-224-5p, a member of the pan-cancer RNC hsa-miR-224-5p_MAGI2-AS3_MBNL2, in EMT program. Our analysis highlights the potential utilities of RNCs in elucidating ncRNA function in cancer, associating with clinical outcomes and discovering novel drug targets or MOA.