Using Attribution Sequence Alignment to Interpret Deep Learning Models for miRNA Binding Site Prediction

• Published in: Biology (Basel, Switzerland) Vol. 12; no. 3; p. 369
• Main Authors: Grešová, Katarína, Vaculík, Ondřej, Alexiou, Panagiotis
• Format: Journal Article
• Language: English
• Published: Switzerland MDPI AG 26-02-2023; MDPI
• Subjects: Back propagation; Binding sites; CLASH; Computer applications; Deep learning; Gene regulation; Genetic transcription; Heuristic; interpretation; Methods; MicroRNA; MicroRNAs; miRNA; miRNA target prediction; Neural networks; Non-coding RNA; Nucleotide sequence; Post-transcription; Prediction models; Protein binding; Protein interaction; Target recognition; visualization; deep learning; visualization; interpretation; miRNA target prediction; CLASH
• ISSN: 2079-7737; 2079-7737
• DOI: 10.3390/biology12030369

MicroRNAs (miRNAs) are small non-coding RNAs that play a central role in the post-transcriptional regulation of biological processes. miRNAs regulate transcripts through direct binding involving the Argonaute protein family. The exact rules of binding are not known, and several in silico miRNA target prediction methods have been developed to date. Deep learning has recently revolutionized miRNA target prediction. However, the higher predictive power comes with a decreased ability to interpret increasingly complex models. Here, we present a novel interpretation technique, called attribution sequence alignment, for miRNA target site prediction models that can interpret such deep learning models on a two-dimensional representation of miRNA and putative target sequence. Our method produces a human readable visual representation of miRNA:target interactions and can be used as a proxy for the further interpretation of biological concepts learned by the neural network. We demonstrate applications of this method in the clustering of experimental data into binding classes, as well as using the method to narrow down predicted miRNA binding sites on long transcript sequences. Importantly, the presented method works with any neural network model trained on a two-dimensional representation of interactions and can be easily extended to further domains such as protein-protein interactions.