Efficient RNA nucleotide encoding enhances the accurate prediction of ac4C modifications

RNA N4-acetylcytidine (ac4C) modification plays a vital role in gene regulation and cellular function. Accurate identification of ac4C sites is essential for elucidating their biological significance. However, existing prediction methods struggle to capture complex sequence patterns, limiting their accuracy. To address this, we propose GO-ac4C, an efficient prediction framework that integrates byte-pair encoding with nucleotide compositional features. GO-ac4C employs dynamic byte-pair encoding to learn optimal subsequence representations and enhances them with compositional features to effectively capture key motifs in RNA sequences. Experimental results demonstrate that GO-ac4C significantly outperforms state-of-the-art methods across multiple evaluation metrics and offers new insights into the mechanisms of RNA modification.