Geng Integrity Report — Qin et al., "TRIM56 Promotes White Adipose Tissue Browning to Attenuate Obesity by Degrading TLE3", Adv. Sci. (DOI: 10.1002/advs.202414073)

Academic fraud report · Geng Detector

Summary

Verdict: Questionable (yellow flag), based on text-only analysis. No pixel-level image forensics was performed because source figures were not provided. The strongest concerns are textual: Figure 3 legend contains duplicated and contradictory panel labels (L is assigned two meanings; M and N are redefined within the same legend), indicating patchwork assembly without global proofreading. Figure 4G mislabels an eWAT group as AAV-TRIM56-iWAT, a classic half-edited copy-paste artifact in the adipose-tissue cell-size statistics. The Abstract contains the typo "subcutaneousl" and the Discussion has "inactivatiion" and "chornic". Lipid-droplet counts in Figure 2F differ roughly twofold (n=79 vs. n=162), which is biologically plausible after browning but warrants threshold-setting records to rule out cherry-picking. Timeline, reagents, and animal-model durations are internally consistent. Image reuse/manipulation (Western blots, histology) cannot be judged and is the main outstanding risk.

Verdict

Questionable (🟡). Based solely on the supplied manuscript text and figure legends, multiple textual inconsistencies and sloppy copy-paste artifacts were detected, but no direct evidence of data fabrication was found. Image-level forensic analysis was not possible; Western blot and microscopy panels in Figures 1B/D, 2E, and 6G remain the highest-priority items for follow-up.

Key findings

Figure 4G legend mislabel (iWAT vs. eWAT): the last group is described as eWAT cell-size measurements but is labeled AAV-TRIM56-iWAT:n=176 cells, inconsistent with the three preceding correctly-labeled groups. Sample sizes: AAV-NC-iWAT:n=101; AAV-TRIM56-iWAT:n=180; AAV-NC-eWAT:n=118; AAV-TRIM56-iWAT:n=176.
Figure 3 legend logic collapse: within a single legend, panel L is first defined as in vitro glycerol release from iWAT strips and then redefined as a preconditioning note for an in vivo experiment; panels M and N are each redefined a second time (serum glycerol; immunoblot analysis), indicating unchecked copy-paste between draft versions.
Proofreading failures in a top-tier journal: Abstract contains "subcutaneousl" (should be "subcutaneous"); Discussion contains "inactivatiion" and "chornic".
Lipid-droplet count asymmetry in Figure 2F: Ad-NC:n=79 vs. Ad-TRIM56:n=162. While browning-induced multilocular droplet formation can increase counts, the roughly 2× difference is sensitive to ImageJ threshold settings and should be cross-checked against the original analysis parameters.
No image forensics performed: Western blot and gel panels, plus histology images, were not analyzed pixel-by-pixel; this is the principal unresolved risk.

Evidence highlights

DOI: 10.1002/advs.202414073
Journal: Advanced Science (Adv. Sci.), published online 10 February 2025; received 31 October 2024; revised 28 January 2025 (≈3.9 months end-to-end).
Authors: Haojie Qin, Yi Zhong, Jinhui Huang, Yanli Miao, Meng Du, Kai Huang.
Figure 4G cell counts: 101 / 180 / 118 / 176 cells across four groups; final group label inconsistent with tissue.
Figure 2F droplet counts: 79 (Ad-NC) vs. 162 (Ad-TRIM56).
Instruments cited: Seahorse XF24 (Agilent), Hitachi HT7800 TEM, GraphPad Prism v.9.0 — all pre-2024, no anachronisms.
Animal model: 8-week-old mice, 2-week acclimation, 12-week HFD; ≈4–5-month total window, internally consistent.

Notes

All image-based claims (Western blot splicing, rotation/reuse, background-pattern matching) require the original high-resolution figures; conclusions on data integrity cannot be finalized until those are examined.
The textual errors are individually minor but their density in figure legends — the most heavily scrutinized section of a biology paper — raises concerns about editorial and peer-review thoroughness rather than outright fraud.
Recommended next steps: obtain the publisher PDF; perform image-level forensics on Figures 1B/D, 2E, and 6G; request Figure 4G raw images and ImageJ threshold logs from the authors; consider a PubPeer comment.