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

Area of Science:

• Biotechnology
• Computational Biology
• Oncology

Background:

• Anti-angiogenic peptides (AAPs) are crucial for inhibiting tumor growth and metastasis.
• Accurate AAP identification significantly impacts cancer therapeutic efficacy.
• High costs and limitations of experimental screening necessitate advanced computational methods.

Purpose of the Study:

• To develop a high-performance deep learning model for predicting anti-angiogenic peptides (AAPs).
• To address the performance limitations of existing computational methods for AAP prediction.

Main Methods:

• Utilized a binary tree structure for the DeepTree-AAPred model.
• Employed protein language models (ProtBERT, ESM-2) for generalized 1D and 2D feature extraction.
• Integrated BiLSTM and TextCNN to capture local features and contextual dependencies for AAP prediction.

Main Results:

• DeepTree-AAPred demonstrated superior performance compared to existing computational methods on standard datasets.
• The model effectively fuses extracted features for accurate AAP prediction.
• Experimental results validate the model's potential for practical application.

Conclusions:

• DeepTree-AAPred offers a significant advancement in computational prediction of anti-angiogenic peptides.
• The model shows promise for improving the efficiency and effectiveness of tumor therapy development.
• This deep learning approach provides a viable alternative to expensive experimental screening for AAPs.