DeepTree-AAPred: Binary tree-based deep learning model for anti-angiogenic peptides prediction

Abstract

Anti-angiogenic peptides (AAPs) show important potential in tumor therapy by limiting the growth and metastasis of tumor cells. Accurate prediction of AAPs is of very positive significance for the therapeutic efficacy of tumors. The high cost of wet experiments limits the application of large-scale screening. Existing computational methods, although able to solve the problem of wet experiments, still lack in performance. To this end, a deep learning-based anti-angiogenic peptide prediction model, DeepTree-AAPred, is proposed in this study. The model utilizes a binary tree structure and employs protein language pre-training models ProtBERT and ESM-2 to extract 1D and 2D generalized features. It further captures local features and contextual dependencies using BiLSTM and TextCNN, ultimately fusing the output features for AAPs prediction. Extensive experimental results on standard datasets show that DeepTree-AAPred outperforms existing computational methods, demonstrating its potential for practical application in AAPs tasks.