Vaccine Design by Reverse Vaccinology and Machine Learning

Area of Science:

• Bioinformatics
• Vaccinology
• Machine Learning

Background:

• Reverse vaccinology (RV) is a leading strategy for vaccine development, utilizing bioinformatics to predict antigens from pathogen genomes.
• Vaxign, a pioneering web-based RV tool, filters proteins based on specific criteria.
• The development of Vaxign-ML leverages machine learning and accumulated protective antigen data.

Purpose of the Study:

• To introduce Vaxign-ML, an advanced machine learning-based method for vaccine antigen prediction.
• To evaluate Vaxign-ML's performance against existing open-source RV tools.
• To demonstrate the utility of the Vaxign platform, including Vaxign-ML, for analyzing and selecting vaccine candidates.

Main Methods:

• Vaxign-ML employs extreme gradient boosting for vaccine antigen prediction.
• Performance was assessed using a benchmark dataset.
• The Vaxign platform integrates Vaxign-ML for web-based access and also offers a command-line version.

Main Results:

• Vaxign-ML demonstrated superior performance compared to other open-source RV tools on a benchmark dataset.
• The Vaxign-ML web implementation provides an intuitive user interface.
• Both Vaxign and Vaxign-ML were successfully applied to predict vaccine antigens for SARS-CoV-2 and Brucella.

Conclusions:

• Vaxign-ML represents a significant advancement in machine learning for reverse vaccinology.
• The Vaxign platform offers a comprehensive and accessible solution for vaccine antigen discovery.
• The integrative approach using Vaxign is effective for identifying potential vaccine candidates against various pathogens.