Geng Academic Integrity Report: Image Duplication Concerns in "Sestrin1 Ameliorates Senescence and Mediates the Protective Effects of Exercise Against Hepatic Abnormality in C57BL/6J Mice" (J Cell Physiol, DOI 10.1002/jcp.70241)

Academic fraud report · Geng Detector

Summary

This AI-assisted forensic review of Yang et al. (2026, Journal of Cellular Physiology, DOI: 10.1002/jcp.70241) concludes the paper is highly likely to contain fabricated Western blot imagery (verdict: confirmed). The core finding is systematic background identity across Western blot panels in Figures 2a, 3k, and 5l: panels for different proteins show identical background texture, brightness, and β-tubulin band morphology, which is physically implausible given required membrane stripping and re-probing. Programmatic image forensics detected copy-move duplication in img-000, img-003, and img-007 (70/24/27 pairs of exact matching blocks at constant offsets (32,0), (32,0), (56,0)). Additional suspicious but unconfirmed findings include apparent duplication of Sestrin1/2/3 bands (Figure 1a), near-identical scatter plot patterns across three independent qPCR genes (Figure 1c-e), and Hoechst nuclear staining images appearing reused across genotypes (Figure 5f). Findings 2-4 lack sufficient standalone forensic support and require original uncropped blots, raw microscopy images, and qPCR Ct values. Overall posterior probability ≥99.9% (BF = 4.33e+4), though this assumes conditional independence of evidence and may overstate correlated evidence. Final determination requires institutional investigation.

Verdict

Confirmed (🔴 实锤) — Systematic image duplication across core Western blot figures (Figures 2, 3, 5) is supported by converging visual analysis and programmatic copy-move forensics. Three additional findings remain unconfirmed pending original data.

Paper: Sestrin1 Ameliorates Senescence and Mediates the Protective Effects of Exercise Against Hepatic Abnormality in C57BL/6J Mice
Authors: Xuege Yang, Yuchen Zou, Xiuru Li, Zhili Rong, Yanmei Niu, Li Fu (Tianjin Medical University)
Journal: Journal of Cellular Physiology, 2026
DOI: 10.1002/jcp.70241 (https://doi.org/10.1002/jcp.70241)

Key findings

Finding 1 (Confirmed, 🔴): Systematically identical Western blot backgrounds across panels. Figure 2a (PGC-1α, MFN2, Drp1, SDHB), Figure 3k (cGAS 59 kDa, STING 33 kDa), and Figure 5l (8 proteins from PGC-1α to IL-6) show near-identical background texture, brightness, noise patterns, and β-tubulin band morphology — physically implausible if membranes were genuinely stripped and re-probed. Caption text for Figure 5 "(l–s) Western blot and quantitative analysis of PGC-1α, MFN2, Drp1, cGAS, STING, IL-1β, TNF-α, and IL-6" is confirmed present (initial programmatic miss due to line-break hyphenation). Forensics corroborated by copy-move detections (I1, I3, I11). No methodological basis in the paper supports benign explanations (unified export pipeline, journal compression).
Finding 2 (Unconfirmed, 🟠): Figure 1a — Sestrin1 (~75 kDa), Sestrin2 (~54 kDa), Sestrin3 (~57 kDa) bands show highly similar background grayscale, noise, and lane-edge vertical cuts, suggesting duplication or over-cropping. No programmatic copy-move evidence; requires original membrane images.
Finding 3 (Unconfirmed, 🟠): Figures 1c, 1d, 1e — scatter plots for independent genes Cdkn1a, Cdkn2a, Trp53 show near-identical point arrangements despite the text stating Cdkn1a/Cdkn2a "increased" while Trp53 "remained unchanged." Column-difference statistical check (S1) is suggestive but premise-limited (small n, possible SEM presentation). Requires raw qPCR Ct values.
Finding 4 (Unconfirmed, 🟠): Figure 5f — Hoechst nuclear staining images appear reused across WT and Sestrin1⁻/⁻ genotypes. Forensics found only 4–5 matching block pairs per image (img-004/005/006) plus noise variance inconsistencies (I4–I9); insufficient standalone evidence. Requires original microscopy images.
Finding 5 (Confirmed, 🔴): Anomalies concentrated in the figures carrying the core conclusions (Sestrin1/Drp1/cGAS-STING axis), not randomly distributed — inconsistent with isolated error or generic processing explanations.

Evidence highlights

Copy-move detection: img-000.png — 70 matching block pairs at offset (32,0); img-003.jpg — 24 pairs at (32,0); img-007.jpg — 27 pairs at (56,0); constant offsets across matches (I1, I3, I11).
Bayesian synthesis: posterior probability ≥99.9% (95% credible interval [99.4%, 100%]); Bayes factor BF = 4.33e+4 (decisive); prior fixed at 5%; 9 clues excluded on review (benign explanations or failed test preconditions).
Top evidentiary contributors: Finding 5 composite forensic result (logLR=1.7, 16.3%), I3 (logLR=1.46, 14.1%), I11 (logLR=1.46, 13%), S1 column-difference statistic (logLR=1.36, 12.5%), I1 (logLR=1.46, 12%).
Recommended actions (per report): request original uncropped blots/scans and raw qPCR Ct values from authors; post on PubPeer; notify the journal (Wiley) and Tianjin Medical University research integrity committee.

Notes

Findings 2–4 are primarily visual in nature with weak or absent programmatic corroboration; benign explanations (repeat exposure, uniform cropping, lighting gradients) have no positive support in the manuscript but remain possible.
The synthesized posterior may overstate combined strength because evidence items were treated as conditionally independent (LR version geng_lr_v3, calibration n=11).
This report is AI-assisted and for academic discussion only; final determination of misconduct requires institutional investigation, and authors' reputation rights are respected. Official investigation conclusions take precedence; false positives and false negatives are possible.