A Universal Pterocarpan-Vinyl-Polyphenol Scaffold for Multi-Target Skin Therapeutics: Six-Cycle Bayesian Active Learning Identifies Six Multi-Target Leaders Across 14 Skin-Disease Targets

Correction (new version, 2026-07-21) — citation correction. Correction (2026-07-21) - misattributed calibration figure. Background: two calibration anchors had been conflated across this corpus. The 'Spearman |rho| ~ 0.72' anchor (n=15) was computed against the ligand panel found to be fabricated (2026-07-16 audit) and is void. The separate 'Pearson R = -0.453, n=93' anchor is REAL - it comes from the ChEMBL-derived 93-compound cohort of companion preprint #18 (pilot/cpu_meaningful/chembl_boltz2_calibration.csv, carrying per-record ChEMBL document identifiers) and reproduces on recomputation (Pearson -0.4535, Spearman -0.4582). It was, however, being cited incorrectly. In this record the limitations section attributed 'pearson correlation against ChEMBL ground truth is r = -0.453' to preprint #8. Preprint #8 never reported that figure - its own calibration was the n=15 one since withdrawn for fabricated potency annotations. The figure belongs to preprint #18's real 93-compound cohort and was measured on Boltz-2 affinity_pred_value; the attribution is corrected. No result, ranking or figure value changes; this is a provenance and scope-of-claim correction. Full audit: preprints/_metadata/DANGLING_CITATION_AUDIT_2026_07_21.md; underlying finding: preprints/_metadata/FABRICATED_PANEL_SCOPE_2026_07_16.md. We report the discovery of a pterocarpan-vinyl-polyphenol scaffold family that acts as a universal molecular template across six independent skin-disease verticals (scar regeneration, pigmentation, alopecia, acne, photoaging, fibrosis cross-disease). Six members of this family were identified through six rounds of Bayesian active learning (R9–R14, 4,597 cofold predictions integrated, 14 protein targets, 200-candidate pool size). All six leaders satisfy the multi-target leadership criterion (top-5 in 8+ of 14 targets) and demonstrate paper-tier 10 ns molecular dynamics stability (RMSD < 2.0 Å) on RTX 5090 GPU. The expected-improvement (EI) acquisition function reached saturation at R12 (EI=0.0077), confirmed across three additional cycles (R13 EI=0.0077, R14 EI not greater than R13). The R12_4/R12_11/R14_5 hydroxymethyl/methoxy variants, R12_23 methyl ester (TYR/TYRP1 selective), R13_13 prenyl variant (R11_0 + lipophilicity enhancement), and R11_0 trihydroxy parent constitute a structure–activity relationship rich enough to support multi-vertical formulation strategies. R15/R16 chromanol follow-up separates a systemic-safety fragment path from a topical-optimized analog path; the R15 triple-safe chromanol fragment remained dynamically stable across its top-3 cofold targets in 30 ns MD, while an 18-pair R16 topical chromanol 30 ns matrix across TGFB1, DCT, and TYR remained paper-tier stable, and the TGFB1 top-6 subset remained stable in 60 ns follow-up (max RMSD 1.22 Å, max last-third mean 0.96 Å). We discuss the six-leader family as in silico-validated lead candidates for Recover Korean Medicine Clinic's universal-skin-care vertical and define wet-lab follow-up via Tier 1 contract research organization (CRO).