Sequence determinant and functional relevance of 8-oxoguanine (o8G) RNA modification unveiled from foundation model-based predictor

8-oxoguanine (o⁸G) is a recently recognized oxidative RNA modification implicated in stress responses and RNA integrity regulation. However, computational approaches for accurately predicting o8G sites and interpreting their sequence determinants are lacking. We constructed the first sequence-based predictor for o8G RNA modification. To address this gap, we fine-tuned a series of pre-trained language models—including BERT2Ome, BioBERT, BERT, DNABERT, and RNABERT—for o⁸G site prediction. These models were trained on curated o⁸G peak datasets and demonstrated high predictive accuracy (AUC > 0.85 across models). To interpret sequence features associated with o⁸G, we applied TF-MoDISco and STREME-based motif discovery, supported by TOMTOM motif comparison. Several co-occurring and biologically relevant sequence motifs were identified, highlighting potential regulatory elements of o⁸G modification. A lightweight web platform was developed to facilitate user-friendly sequence input and prediction.