ABSTRACT
MicroRNA (miRNA) is a major type of small RNAs that plays a significant role in the post-transcriptional control of gene expression by targeting messenger RNAs (mRNAs). Technological advances have revealed the function of miRNA in controlling a blanket of biological processes and signaling interactions. miRNA-based regulation is implicated in disease etiology and has been potentially studied for developing therapeutic targets. Thus it is of the utmost importance for the robust prediction of miRNAs and their targets to better understand the function and decipher their regulating functions. Deep learning is a subfield of machine learning methods inspired by artificial neural networks (ANN). By effectively deploying large data sets, deep learning has shown promising results in the field of visual computing and natural language processing. Because of its ability to learn unknown features from raw data, it tends to be a preferred method for various modeling tasks in omics research. In this review, we summarize different deep learning modeling techniques and their existing applications in the analysis of miRNA sequences. Widely used deep learning methods such as deep neural networks (DNN), convolution neural networks (CNN), recurrent neural networks (RNN), and auto encoders have been described. Further, various deep learning approaches and open challenges specific to miRNA modeling tasks including miRNA prediction, precursor detection, and target prediction have been outlined.
