To solve this problem, this study proposes a feature fusion-based prediction model, where the protein pre-trained language models ProtBERT and ESM-2 are used as feature extractors, and the extracted features from both are fused to obtain a more comprehensive and effective feature representation, ...