Ac4CGE: Predicting N4-acetylcytidine (ac4C) RNA Modification Sites in Archaea Using Graph-based Machine Learning Approach

N4-acetylcytidine(ac4C) is one of the highly conserved epigenetic modifications in RNA, possessing the ability to facilitate mRNA expression, enhance its stability, and influence mRNA decoding efficiency, thereby promoting substrate translation. The identification of ac4C in archaea has been considered a laborious and time-intensive endeavor using conventional biological methods. Therefore, it is essential to develop a computational method with high confidence to explore the ac4C modification sites. In this study, we present the first predictor, ac4CDE, which utilizes both sequence-derived and graph-embedding features to predict ac4C modification sites in an archaea dataset. The result achieves remarkable accuracy and performance. The best feature, NCP-ND, achieves Matthew's correlation coefficient of 0.9921, an F1-score of 0.9929, and a mean squared error of 0.0014. These findings underscore the applicability and robustness of our model as a valuable tool for predicting N4-acetylcytidine sites.