Protein Sequence Prediction Based on Feature Combination and Attention Mechanism

Identifying protein functions is crucial for understanding biological mechanisms and advancing life sciences. However, current protein prediction methods fail to fully exploit sequence information, resulting in limitations in prediction accuracy. To address this issue, we propose a protein sequence prediction network (ECPN-HFGF-ATT) that incorporates feature combination and an attention mechanism. The proposed method initially employs a residual network to extract shared features from protein sequences, followed by hierarchical and global feature extraction modules to capture more detailed hierarchical and global sequence characteristics. Next, the prediction results from both feature types are integrated, and the attention mechanism is applied to enhance prediction accuracy. Experimental results demonstrate that the ECPN-HFGF-ATT method achieves superior performance in protein sequence prediction, with macro and micro F1 scores of 95.5% and 99.0%, respectively. The ECPN-HFGF-ATT method effectively integrates hierarchical and global features, enabling rapid and accurate identification of protein sequences. Compared to commonly used methods, this approach significantly enhances prediction accuracy, offering a powerful tool for advancing protein sequence research and applications.