AntiCVP-Deep: Identify anti-coronavirus peptides between different negative datasets based on self-attention and deep learning

Abstract

The disease caused by infection with the SARS-CoV-2 virus is threatening people's health and lives. Antimicrobial peptides (AMPs) with anti-coronavirus (anti-CoV) functions may be potentially effective therapeutics for the treatment of infected coronaviruses. Due to the complexity of experimental methods, it is particularly important to identify anti-CoV peptides by computational techniques. In this paper, we propose a new predictive model, called AntiCVP-Deep. First, six feature extraction methods are used to obtain the original feature vector of antimicrobial peptides, and K-means SMOTE is used to handle imbalance data. Then, the processed balanced sample data is passed through the input gate, forget gate and output gate of a bidirectional long short-term memory network (BILSTM) to select optimal features. Next, the important antimicrobial peptides information is given a higher weight through the self-attention mechanism, which enhances the model's ability to learn features. Finally, the data from self-attention layer are input into fully connected neural network (FCN) to predict anti-CoV peptides. The AUC values of the four datasets all reach above 98%. The geometric mean (GMean) of antivirus, non-AVP, non-AMP and all-Neg recognition on the independent test sets reach 90.05%, 92.63%, 99.46% and 91.24%, respectively. This experimental result shows that the AntiCVP-Deep method is helpful for identifying anti-CoV peptides and has better predictive performance compared to other existing predictive models.