Predicting miRNA Targets at the isomiR Level

Understanding how microRNAs (miRNAs) regulate gene expression through target binding is critical for comprehending gene regulation mechanisms. Despite many experimental and computational studies on miRNAs and the availability of various tools for identifying their targets, accurately pinpointing these targets remains a significant challenge. The presence of various miRNA isoforms, known as isomiRs, further complicates this task due to their differing abundances in tissues. To address this challenge, we introduce DMISO, the first and currently only computational tool specifically designed for miRNA and isomiR target identification. DMISO employs deep learning to predict target sites for both isomiRs and miRNAs using a model trained on CLASH (cross-linking, ligation, and sequencing of hybrids) data. In this chapter, we start with a background introduction to isomiR and miRNA target prediction. We then outline the data available for identifying isomiR and miRNA targets and provide a detailed overview of DMISO, including its key features, performance compared to existing tools, and instructions for its installation and use. Finally, we discuss potential applications of the DMISO tool. This chapter aims to provide a foundational understanding for further research on isomiRs and their roles in gene regulation.