An epitope-boundary audit with two-predictor HLA-I landscape mapping and consensus validation of YKEFGA[S/T]-overlapping HBc/HBe N-terminal windows

Paper D of the HBV-VISLM_sys3 series — an in silico, hypothesis-generating study of whether the conserved hepatitis B virus core/precore (HBc/HBe) N-terminal motif YKEFGA[S/T] (residues 6–12) is immunologically visible, and which motif-overlapping peptide registers should be advanced to wet-lab validation. Background & Aim. The HBc/HBe N-terminal YKEFGA[S/T] motif is highly conserved and capsid-associated, but whether it marks an immune-visible T-cell target — and at which peptide unit — remains unclear. We performed an in silico epitope-boundary audit, HLA-I landscape mapping, and consensus prioritization of motif-overlapping N-terminal windows. Methods. Indexed IEDB records were audited for exact and containing epitopes. Motif-overlapping 8–11-mers from both S12 and T12 motif states were screened with NetMHCpan-4.1 across 27 HLA class-I alleles and re-scored with MHCflurry-2.2.1, generating a 2,112-cell register/length concordance map. Priority candidates were further cross-validated with MixMHCpred-3.0 and NetMHCcons-1.1 to form a four-predictor binding consensus, with NetMHCstabpan-1.0 used as a separate peptide–MHC complex-stability axis. Per-position conservation across 1,827 HBc/precore sequences, capsid solvent accessibility using PDB 2G33, and human-proteome self-similarity by exact matching and BLASTp/PAM30 near-identity were integrated. Results. The exact motif-level queries had no characterized positive epitope, and S-genotype short windows assayed on HLA-A*02:01 were binding-negative and predictor-weak. Two-predictor landscape mapping resolved the upstream HBc/HBe N-terminal region, distinct from the classical HBc18–27 CTL epitope, into a K7-anchored CORE family (KEFGA[S/T]VEL on HLA-B*40:01 and HLA-B*39:01) and a D4-anchored FULL-MOTIF family (DPYKEFGA[S/T]V on HLA-B*51:01 and HLA-B*54:01). On the priority panel, CORE KEFGA[S/T]VEL/HLA-B*40:01 and FULL-MOTIF DPYKEFGA[S/T]V/HLA-B*51:01 were called binders by all four predictors and were predicted to form stable complexes. In contrast, the HLA-A*02:01-negative controls remained weak across all four predictors. Adding the buried N-terminal residue Y6 worsened predicted presentation in 44/44 paired allele×variant comparisons; accordingly, the supported CORE register is the K7-anchored 9-mer rather than the Y6-extended 10-mer. Priority windows were conserved and lacked exact ≥8-mer identity or ≤1-mismatch ≥8-mer human-proteome near-identity. Conclusions. YKEFGA[S/T] is not a validated minimal T-cell epitope. It functions as a structural and coordinate anchor that defines a consensus-supported, wet-lab-ready, allele-matched HLA-I candidate-window landscape requiring experimental validation. All positive HLA-I results are computational predictions; predicted binders are reported as HLA-binding candidates only, never as validated epitopes, and no candidate has been experimentally tested. Deposit contents. This record contains the manuscript (DOCX; text plus Tables 1–7), a separate figure-embedded PDF reading copy, Figures 1–5 and Supplementary Figure S1 as deposited image/vector files, Supplementary Tables S1–S8 (CSV), Supplementary Methods S1–S11, reproducible analysis scripts, and reference files in BibTeX and RIS formats. Analyses are CPU-compatible and were run primarily in Python 3.12. Tools used include NetMHCpan-4.1, MHCflurry-2.2.1, MixMHCpred-3.0, NetMHCcons-1.1, NetMHCstabpan-1.0, NetMHCIIpan, NCBI BLAST+, and FreeSASA. NetMHCcons and NetMHCstabpan were accessed through the IEDB Analysis Resource API; MixMHCpred was run locally. Related records (HBV-VISLM_sys3 series).Paper A — 10.5281/zenodo.20529386Paper B — 10.5281/zenodo.20571631Paper C — 10.5281/zenodo.20733773 License: Creative Commons Attribution 4.0 International (CC BY 4.0).