Researchers develop AI tool to map microRNA binding sites on circular RNAs
Circular RNAs (circRNAs) are loop-shaped molecules that can influence gene regulation through interactions with microRNAs (miRNAs). This matters because circRNAs can exist as different full-length isoforms, and their back-splice junction (BSJ) connects the two ends of the RNA. Treating a circular RNA like a linear transcript can therefore overlook sequence information near this junction.
Researchers at Pusan National University have created an artificial intelligence tool, named circMAC, to identify potential microRNA binding sites within circular RNAs. Circular RNAs, or circRNAs, are a unique class of RNA molecules with a looped structure that can affect gene regulation via interactions with microRNAs. Traditional methods of analyzing these molecules often overlook the sequence information near the back-splice junction, which connects the two ends of the RNA.
The circMAC framework addresses this issue by taking a full-length circRNA isoform and a mature miRNA as inputs and predicting the binding probability for every nucleotide within the circRNA. Its predictive power stems from a combination of three sequence-modeling techniques: attention, Mamba models, and a one-dimensional convolutional neural network.
This combination preserves the distinctive circular sequence by modeling information from across the circRNA, sequential dependencies, and short local patterns associated with binding sites.
The model also incorporates cross-attention to account for the specific miRNA being analyzed, allowing each circRNA nucleotide to be evaluated in its miRNA-specific context. Additionally, circular padding in the convolutional branch enables the consideration of short local patterns spanning the back-splice junction continuously.
The researchers validated circMAC using full-length circRNA isoforms with AGO-CLIP binding-site annotations, achieving high performance metrics, including an AUROC of 0.902 and an AUPRC of 0.545 in a pair-disjoint test.
Despite its promising results, it is important to note that circMAC's predictions should be viewed as potential candidate regions rather than established functional interactions or clinical applications. The tool is designed to prioritize candidate binding regions for follow-up experimental validation, providing more focused areas for researchers to investigate.
Overall, circMAC represents a significant step forward in the computational study of RNA-mediated regulation, offering a machine-learning framework to prioritize candidate circRNA-miRNA binding regions with increased precision.
Written by urgent.news from Phys.org's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
