DPNN-ac4C

RNA ac4C modification is a highly conserved RNA modification in prokaryotes and eukaryotes, which plays an important role in translation efficiency, mRNA stability, and gene expression regulation. However, traditional ac4C detection methods rely on complex biotechnological approaches, which are time-consuming and costly. In this study, we propose a dual-path neural network with self-attention mechanism (DPNN-ac4C) to efficiently identify ac4C sites in mRNA. The model combines embedding modules, bidirectional GRU networks, convolutional neural networks, and self-attention mechanism, which effectively extract and utilize both local and global features of RNA sequences. Through evaluation on an independent test set, the DPNN-ac4C model demonstrates excellent performance, with an area under the receiver operating characteristic curve (AUROC) of 91.03%, accuracy (ACC) of 82.78%, Matthews correlation coefficient (MCC) of 65.78%, and specificity (SPE) of 84.78%. Furthermore, the model maintains a high level of accuracy in robustness testing under the Fast Gradient Method (FGM) attack, demonstrating its reliability in practical applications. Compared to existing models, DPNN-ac4C shows better performance on multiple evaluation metrics, providing a new and efficient deep learning approach for the identification of RNA ac4C sites. The model code and dataset can be obtained from https://github.com/shock1ng/DPNN-ac4C.