LEVERAGING MACHINE LEARNING FOR THE IDENTIFICATION AND FUNCTIONAL CHARACTERIZATION OF NON-CODING RNAS: ADVANCING GENOMIC INSIGHTS AND THERAPEUTIC APPLICATIONS

Abstract

The existence of longy noncoding RNAs and their functional relation to human genes remain rather obscure at present because ncRNAs are structurally complex and are barely annotated in genomic databases. An important resource for the development of new definitions of the biological importance and potential application of ncRNAs in therapy is ML which has made it possible to accurately classify and predict their functionality. Therefore, this study evaluates Supervised learning, deep learning and pathway analysis to identify and categorize ncRNAs grounded on sequence, structural, and expression characteristics. The feature selection was performed together with SVMs, RFs, CNNs, and RNNs and a large dataset of ncRNA sequences was analyzed. On its part, CNNs had the highest accuracy in the prediction when conducting research on the subject. The link of major ncRNAs to the disease mechanism was further supported by functional annotation and pathway enrichment analysis where the implicated ncRNAs are in active biological processes including the PI3K-Akt and Wnt signaling pathway. Additionally, it was also indicated that ncRNAs can serve as biomarkers for cancer, heart disease, and neurological disorders in the light of the machine learning-based ncRNA-disease association studies. The outcomes reveal the recognition that machine learning might improve the exact diagnosis and specialized treatment plans for fresh NC RNA approaches of treatment. To enrich the ncRNA annotation and expound their function in the future, much attention should be paid to the machine learning based method with interpretability and the multiomics strategy.