Machine learning approaches for plant miRNA prediction: Challenges, advancements, and future directions

Abstract

MicroRNA (miRNA) is an important regulator of gene expression in plants that functions to regulate plant development and growth. Signature sequences and secondary structures that are formed as a result of RNA processing during miRNA biogenesis can be used to identify unique miRNAs and has led to the development of multiple bioinformatic tools for miRNA prediction. Artificial intelligence (AI), particularly machine learning (ML), has enabled significant progress to be made in various tasks such as image recognition, regulatory element identification in genomes, and gene expression prediction. AI-based tools have also been used to predict plant miRNAs and are as accurate as pattern-based tools. In this review, we focus on the use of AI to predict plant miRNAs. We demonstrate how to establish a predictive ML model to predict miRNAs and describe the potential pitfalls at each step. We then introduce tools based on ML that have been published and systematically assess their performance. Finally, we discuss how new AI tools should be developed for plant miRNA researchers.