DCN Analysis Identifies Co-occurring Epitranscriptomic Alterations Across Dozens of Cancer Subtypes

Cancer encompasses a wide array of pathology stemming from complex genetic and epigenetic changes, with N6-methyladenosine (m6A) RNA methylation being the most identified type of modification involved in the initiation or progression causing cancer. This dissertation will expand the m6A epitranscriptome research of 12 cancers and their matched noncancer samples into depth with the integration of a combined total of 167 profiling datasets. In assessing any mechanisms involving m6A, the method we suggest conducting our analysis combines protozoological and generalist methods. Here, we will also develop 12 different deep-learning models to assist with identifying the cancerous m6A signatures. Additionally, in any of our models, and attempts to statistically differentiate tumor mass from healthy samples, we also include RNA sequencing type features and genomic labelling data to help optimize accuracy. Following our example, we also used the newly discovered features to investigate type-specific molecular subtype patterns across different cancers by developing a pan-cancer m6A model, which led to findings of an evolutionary conserved epitranscriptomic profile associated with damaging disruption of RNA hybridization and splice regulatory mechanisms. Together the cancer type specific and cross cancer research of the present study provides a new complementary perspective on the role m6A modifications have in cancer pathogenesis. The results present a new understanding of the molecular mechanisms of cancer progression which aids both the development of new treatments and potentially other cancer biomarkers based on the conserved features identified.