Alignment-anchored validation reveals conserved motifs in hepatitis B virus core protein, suggesting immunological implications.
Preprint — version 1.0 (not peer-reviewed). Computational virology / bioinformatics. This record is the preprint of “Alignment-Anchored Validation of a Conserved YKEFGA[S/T] Motif and S/T12 Motif-State Polymorphism in Hepatitis B Virus Core Protein.” It reports an alignment-anchored, leakage-controlled population analysis that defines and validates a conserved seven-residue N-terminal motif of the hepatitis B virus core protein (HBcAg) — YKEFGA[S/T] at amino acids 6–12, corresponding to nucleotides 1918–1938 on the NC_003977.2 reference — together with a single-residue serine/threonine polymorphism at position 12. The motif was not taken from the literature as a named motif: it was nominated during earlier project-internal, model-guided and therapeutic-target–oriented exploration and is here defined and validated at the population level, with trained CNN/Transformer interpretability from the companion study used only as weak cross-validation rather than as a discovery claim. Key findings The motif coordinate (NT 1918–1938) was verified by two-step in-frame translation and is conserved at ≥96.3% exact match in every genotype across a deduplicated 1,828-sequence full-genome cohort (genotypes A–F). The position-12 polymorphism is motif-state-structured rather than genotype-B-specific: an S12-dominant group (B, C, F; ≥96.1% S12) and a T12-dominant group (A, D, E; ≥75.1% T12), with non-overlapping Wilson 95% confidence intervals between the two groups. These are motif-state groupings at HBcAg position 12, not genome-wide phylogenetic clades. Trained-classifier interpretability gives only weak, above-baseline Transformer first-layer attention (1.02–1.23× the uniform expectation) and a negative CNN saliency result, supporting population validation rather than de novo deep-learning discovery. The motif does not overlap any published capsid-assembly-modulator (HAP/CpAM) pocket, dimer–dimer-interface residue or modulator-resistance position (all at or C-terminal to residue 25); it coincides with the N-terminal HBcAg CD4+ T-helper antigen context, indicating an immunological rather than a capsid-drug-pocket relevance. All claims are computational. Experimental validation of differential HLA-restricted presentation of S12 versus T12 variants, and structural assessment of the two dimer variants, remain future work. Contents of this record. A reproducibility and provenance archive (HBV-VISLM_paperB-zenodo.zip) containing the manuscript (image-free DOCX and PDF, plus references.bib/.ris), the four main figures (PNG and PDF), the per-genotype motif-scan and attention/saliency result bundles, the promoted source tables (Tables 1–4), the text-only supplementary materials (Supplementary Methods S1–S2 and Supplementary Tables S1–S2), a data-provenance pointer, and a SHA-256 manifest. A separate figure-embedded manuscript PDF is included on the same record for convenient inline reading. Companion study. The deduplicated 1,828-sequence cohort, the reference-anchored MAFFT alignment, and the deterministic CNN/Transformer model framework are reused from the companion deposit: Park YM, Complementary CNN and Transformer architectures for HBV genotype classification and genotype-boundary interpretation, Zenodo, doi:10.5281/zenodo.20529386. The large alignment data are not duplicated here; their source files and SHA-256 hashes are listed in the archive (data/DATA_README.md). Trained model weights are not redeposited and are regenerable from the companion-study notebooks or available on request. Paper B archives the motif-specific coordinate, frequency, attention/saliency and source-table outputs used for all manuscript claims. Use of generative AI. Large language models were used as computational and writing aids for code drafting, manuscript editing and reference verification; all analyses, interpretations and final text are the author’s own. No generative-AI-created image content appears within the figures. License: Creative Commons Attribution 4.0 International (CC BY 4.0).Funding: none.Competing interests: none declared. Cite as: Park YM. Alignment-Anchored Validation of a Conserved YKEFGA[S/T] Motif and S/T12 Motif-State Polymorphism in Hepatitis B Virus Core Protein. Zenodo; 2026. doi:10.5281/zenodo.20571631.
No takes yet. Share an insight, caveat, or question.
Young Min Park (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: