Introduction to IPs-GRUAtt

The deep learning model framework based on Bi-GRU and attention mechanism to predict the phosphorylation sites of SARS-CoV-2 infection. Given a protein sequence, it is encoded by an embedding layer and then fed into the Bi-GRU structure. After Bi-GRU, we add dropout to prevent model overfitting, and also use an attention mechanism layer to capture the position information of the protein sequence. Finally, the input protein sequences are discriminated by a fully connected layer. The experimental results show that IPs-GRUAtt surpassed the existing methods for identifying phosphorylation sites.