Abstract: Protein residue interaction prediction plays an important role in constructing protein tertiary structure, and many prediction methods based on deep learning have been proposed. The interaction of the residue pairs is not only determined by the characteristics of the residue pairs themselves, but also affected by all other residue pairs. However, the convolutional neural networks used in most depth models pay more attention to the extraction of local features, while ignoring the global factors of the residue pair interaction. To solve this problem, a prediction model based on attention mechanism is proposed, which can take into account both local and global properties of residue pairs. Experimental results show that the proposed model outperforms other models.