Recent Advances in Machine Learning-Based Models for Prediction of Antiviral Peptides.

Area of Science:

• Biochemistry and Molecular Biology
• Computational Biology
• Drug Discovery

Background:

• Viral infections, including COVID-19, HIV, and hepatitis, pose significant global health challenges.
• Antiviral peptides (AVPs) are increasingly utilized in therapeutic drug design to combat these infections.
• Accurate identification of AVPs is critical for advancing pharmaceutical research and development.

Purpose of the Study:

• To conduct a comprehensive review of existing antiviral peptide (AVP) identification methods.
• To analyze the datasets, feature representation, classification algorithms, and performance evaluation metrics used in AVP prediction.
• To identify limitations in current AVP identification strategies and propose future research directions.

Main Methods:

• Systematic literature review of experimental and computational AVP identification approaches.
• Analysis of applied datasets, feature encoding techniques, and machine learning classifiers.
• Evaluation of performance metrics and comparison of different prediction models.

Main Results:

• Existing AVP identification methods have limitations in accuracy and efficiency.
• A detailed overview of various feature representation and classification algorithms is provided.
• The pros and cons of commonly used classifiers for AVP prediction are discussed.

Conclusions:

• More accurate predictors are needed to enhance the identification of antiviral peptides (AVPs).
• Future research should focus on advanced feature encoding, optimization, and classification techniques.
• Improved AVP prediction methods will significantly benefit drug design and the pharmaceutical industry.