Deep learning-based microRNA target prediction using experimental negative data

MicroRNAs (miRNAs) are small non-coding RNA molecules that control the function of their target messenger RNAs (mRNAs). As miRNAs regulate their target genes by binding them, investigating miRNAs is important to understand various biological processes. Although there exists a deluge of computational tools, reducing the number of false positives (i.e., non-functional targets) has been challenging. To solve this problem, this paper proposes an end-to-end machine learning framework for functional miRNA target prediction. The proposed approach exploits one-dimensional convolutional neural networks (CNNs) based on sequence-to-sequence interaction learning framework and utilize experimental negative data instead of mock ones. As the result, the proposed approach achieved 10% increase in F-measure compared to the existing alternatives.