AI4AVP: an antiviral peptides predictor in deep learning approach with generative adversarial network data...

Area of Science:

• Computational biology
• Drug discovery
• Bioinformatics

Background:

• Antiviral peptides (AVPs) show promise for treating viral diseases.
• Advances in deep learning enable in silico peptide drug design.

Purpose of the Study:

• To develop a deep learning model for predicting antiviral peptides.
• To create a user-friendly web server for AVP research.

Main Methods:

• Collected and curated AVPs.
• Developed a sequence-based binary classifier using deep learning.
• Employed a generative adversarial network (GAN) for data augmentation.
• Utilized a convolutional neural network (CNN) for AVP prediction.

Main Results:

• The developed deep learning classifier outperformed existing state-of-the-art methods.
• The AI4AVP web server provides a valuable resource for the research community.
• Trained classifiers and code are publicly available.

Conclusions:

• Deep learning models can effectively predict antiviral peptides.
• AI4AVP facilitates in silico AVP drug design.
• The AI4AVP server and associated code support further research in antiviral peptide discovery.