The critical role of artificial intelligence and bioinformatics in accelerating peptide-based vaccine discovery for tackling global infectious diseases

Peptide-based vaccines, enabled by bioinformatics and machine learning (ML), have emerged as one of the most promising approaches for rapid, safe, and cost-effective vaccine design against infectious diseases. Unlike conventional approaches that depend heavily on whole-pathogen cultures or recombinant protein expression, peptide vaccines can be designed in silico and synthesized quickly. Rational and targeted in silico approaches for the discovery of peptide-based vaccine candidates include B-cell and T-cell epitope prediction, immunogenicity, antigenicity, allergenicity, autoimmunity, population coverage, sequence conservation, molecular docking, molecular dynamics simulation, in silico cloning, and immunological simulation analyses. The combination of these comprehensive computational methods can effectively generate high-quality vaccine candidates for subsequent validation via in vitro and in vivo experiments. This review contextualizes the historical trajectory of peptide-based vaccinology, from early linear epitope discoveries in the 1960s to multi-epitope constructs and clinically tested candidates such as UB-612 and PepGNP-Covid19. It examines critical challenges in immunoinformatics, including performance gaps in epitope prediction tools, complexities in human leucocyte antigen (HLA) mapping, and the need for extensive manual intervention in pipelines. Artificial intelligence-driven approaches, spanning deep learning, and interpretable ML, are positioned to transform epitope prediction, reduce human error, and standardize reproducibility. These advances have the potential to support global outbreak response targets such as the Coalition for Epidemic Preparedness Innovations (CEPI) 100 Days Mission and the World Health Organization (WHO) Research and Development (R&D) Blueprint. However, their performance remains constrained by data quality, dataset imbalance, limited benchmark standardization, and persistent underrepresentation of many HLA alleles and population groups.