Uncropped Film Scans and Analysis Data Addressing PubPeer Query on JNC 2009_Socodato et al.

In response to the comment flagging similarities in Figs 2, 3, and 5 (red and yellow rectangles), we retrieved the legacy uncropped films and ran a lane‑agnostic forensic pipeline on the full band regions (no lane selection). The PubPeer PDF page with the marked rectangles is available here for context. Llane‑agnostic, FDR‑controlled analysis. We compared whole‑strip images using three independent metric families, each evaluated with permutation/KS tests and Benjamini–Hochberg FDR across families per pair: Spectral distance of the radial power spectrum (phase‑randomized null) → p, q, z. Cross‑duplication via dense tile descriptors (auto‑scaled tiles/stride) → p, q, z. Stationarity via KS on residual‑variance maps (illumination/heteroscedasticity proxy) → p, q, D. Decision rule (pre‑specified): a pair is flagged only if ≥2 families have q<0.05 and medium+ effect (|z|≥2 for spectral/cross‑dup; D≥0.2 for stationarity). Otherwise: inconclusive (i.e., not flagged for duplication). Results for the rectangles highlighted on PubPeer. Red rectangles — Fig. 2a vs Fig. 3a (pCREB): spectral q=0.743 (z=−0.696); cross‑dup q=0.704 (z=−0.060); stationarity q≈4.84×10⁻²⁴⁸, D=0.373 → decision: inconclusive. Red rectangles — Fig. 2a vs Fig. 3a (pERK): spectral q=0.704 (z=−0.283); cross‑dup q=0.704 (z=−0.012); stationarity q≈9.36×10⁻³⁷, D=0.144 → decision: inconclusive. Yellow rectangles — Fig. 3b vs Fig. 5a (pCREB): spectral q=0.557 (z=−0.198); cross‑dup q=0.557 (z=+0.088); stationarity q≈1.10×10⁻⁶¹, D=0.186 → decision: inconclusive. Interpretation: the duplication‑focused families (spectral distance and cross‑duplication) are non‑significant after FDRfor all three comparisons; the stationarity family indicates differences in background/illumination (e.g., crease/gradient), which is consistent with different exposures, not re‑use. Additional whole‑strip cross‑checks (no lanes). 3b pCREB vs 5a pCREB: SSIM (after global alignment) ≈ 0.62; phase‑correlation response ≈ 0.025; 0 geometric inliers on ORB+RANSAC — inconsistent with duplication. 2a vs 3a (pERK / pCREB): SSIM ≈ 0.54/0.45; phase‑correlation ≈ 0.060/0.049; negligible geometric consensus (0–13 inliers across hundreds of keypoints). (All figures and metrics computed on the full band region without lane picking.) Conclusion. Across the red (2a↔3a pERK/pCREB) and yellow (3b↔5a pCREB) rectangles, a lane‑agnostic, FDR‑controlled analysis finds no evidence of panel re‑use. The only strong signals are stationarity differences, which reflect different background/illumination completely under normality given different/independent scanned films.