Deep Learning Techniques for pre-miRNA prediction: Current Challenges and Future Directions

Micro RNAs (miRNAs) are small non coding RNA sequences that incriminated in numerous biological, anatomical processes and diseases. miRNAs are generally consisting of 20~23 nucleotides. miRNAs govern the post transcriptional expression of genes in both normal and disease condition of the cell. Thus, identification of miRNAs in the RNA sequence becomes crucial for understanding disease pathogenesis. Because of its short length, it is of-ten difficult to distinguish a mature miRNA from other non-coding RNA sequences. Thus, researchers aim at identifying precursor miRNAs (pre-miRNAs). Deep Learning is a part of broader family of machine learning methods which is inspired by artificial neural networks (ANN). By effective-ly deploying large data sets, deep learning has shown promising results in the field of visual computing and natural language processing. Now it tends to be a preferred method for various modeling tasks in omics research. In this re-view different emerging deep learning architectures and their applications in pre-miRNA prediction are outlined. A concise description of each study and various open challenges specific to pre-miRNA discovery has been present-ed. This study provides a research direction and perspective on future developments on emerging deep learning models for precise identification of miRNAs.