Background/Aim: A major barrier in vaccine development is the limited population-level effectiveness of epitopes that demonstrate strong immunogenicity in experimental assays but fail under real-world HLA heterogeneity and T-cell receptor (TCR) recognition constraints. This challenge is especially pronounced in the genetically diverse Asia-Pacific region. We aimed to develop a bioinformatics framework that jointly integrates experimental immunogenicity, Asia-Pacific HLA population coverage, and TCR recognizability to prioritize vaccine epitopes with superior translational robustness.
Methods: We integrated validated T-cell epitopes with quantitative assay evidence from the Immune Epitope Database (IEDB), antigen-specific TCRβ sequences from VDJdb, and Asia-Pacific HLA allele frequency proxies derived from the 1000 Genomes Project. A heterogeneous epitope–HLA–TCR graph encoded restriction patterns, population presentation, and public TCR motif density. An APAC-Robust Epitope Score combined Bayesian-shrunk assay responder probabilities, expected Asia-Pacific population coverage, and TCR recognizability measured by motif redundancy and clonotype recurrence. To minimize bias from uneven assay density, epitope-level features were regularized using empirical Bayes priors, and sensitivity analyses confirmed score stability after excluding low-frequency HLA alleles and single-study TCR motifs. Graph-based link prediction was evaluated using pathogen-stratified 5-fold cross-validation with isotonic calibration and 1,000 bootstrap resamples.
Results: Across 21,418 validated epitopes and 62,105 antigen-specific TCRs, triangulated scoring outperformed assay-only prioritization in identifying high-confidence CD8⁺ epitopes (AUROC 0.81, 95% CI 0.79–0.83 vs 0.69, 95% CI 0.66–0.72; p<0.001). The top 50 epitopes achieved higher expected Asia-Pacific population coverage (median 83.9%, IQR 77.4–89.1) than HLA-binding–based selection (60.8%, IQR 51.6–71.2; p<1×10⁻⁶), alongside a 39% increase in public TCR motif density (95% CI 30–48). Sensitivity analyses demonstrated minimal performance attenuation after removal of rare HLA alleles (ΔAUROC −0.01) and study-specific TCRs, supporting robustness against sampling imbalance and dataset-specific annotation noise. A minimal 12-epitope panel maintained ≥80% coverage across all Asia-Pacific subpopulations with stable calibration (Brier score 0.112, 95% CI 0.105–0.119). Nine conserved public TCR motifs were identified as candidate post-vaccination immune monitoring readouts. Subgroup analyses showed consistent performance across viral and bacterial antigens, with AUROC ranging 0.78–0.83 and no significant interaction by pathogen class (interaction p=0.41), supporting broad applicability. Decision-curve analysis showed higher net benefit across plausible risk thresholds, indicating epitope selection utility for vaccine design under deployment scenarios.
Conclusions: Epitope–HLA–TCR triangulation enables population-aware antigen prioritization that overcomes key limitations of conventional immunogenicity screening. This approach delivers compact, Asia-Pacific–robust epitope panels with integrated immune readouts, supporting next-generation vaccine and immunotherapy development.