Generative Adversarial Network-Based Augmentation With Noval 2-Step Authentication for Anti-Coronavirus Peptide...

Area of Science:

• Biotechnology
• Computational Biology
• Drug Discovery

Background:

• Viral diseases present a persistent global health threat, necessitating novel therapeutic strategies.
• Antiviral peptides offer a promising avenue due to their efficacy and favorable safety profiles.
• Current methods for identifying and predicting antiviral peptide activity require enhancement.

Purpose of the Study:

• To develop and validate a novel approach for augmenting antiviral peptide data using generative adversarial networks (GANs).
• To enhance the accuracy of antiviral peptide activity prediction through a two-step authentication process for synthetic peptides.
• To evaluate the performance of deep learning models in classifying antiviral peptides using augmented datasets.

Main Methods:

• A generative adversarial network (GAN) was utilized for antiviral peptide data augmentation.
• A two-step authentication process involving NCBI-BLAST and physicochemical property comparison was implemented for synthetic peptides.
• Five deep learning models, including a 1-D convolution neural network, were employed for classification tasks.

Main Results:

• The 1-D convolution neural network, utilizing augmented peptide data, achieved superior performance over other models.
• The model attained a mean classification accuracy of 95.41%, an AUC of 0.95, and an MCC of 0.90.
• Authenticated peptide augmentation significantly improved prediction outcomes compared to non-augmented data.

Conclusions:

• The proposed method effectively enhances antiviral peptide prediction accuracy.
• Generative adversarial networks combined with a robust authentication process show significant potential for drug discovery.
• The developed model demonstrates high efficacy in predicting antiviral peptide activity, offering a valuable tool for combating viral infections.