Investigation Report: 'Neuronal Sensitization Drives Bone Regeneration via Energy Metabolism and BMP2-ErbB1 Signaling' (Adv. Funct. Mater., DOI: 10.1002/adfm.76607)

Academic fraud report · Geng Detector

Summary

This report assesses multiple data-integrity concerns in Dai et al. (Advanced Functional Materials, DOI: 10.1002/adfm.76607). The most serious finding is an implausible ZDOCK docking score of 13553.6 for ErbB1 in Table 1, far exceeding the typical 1000–3000 range of the rigid-body FFT-based algorithm and inconsistent with all other receptors in the table (1528–1967), suggesting fabrication or technical illiteracy. A second concern is identical binding energies of exactly −17.7 kcal/mol for ErbB1 and ErbB2, statistically improbable and indicative of copy-paste. Third, the Methods claim ICP-OES quantification but cite a PerkinElmer PinAAcle 900T, which is an atomic absorption spectrometer (AAS), not an ICP-OES instrument, indicating methodological error or copy-paste from an unrelated protocol. Fourth, the STRING network p-value of 0.0173 for a 37-node network is implausibly high for the platform's enrichment algorithm. Image-based analyses were not performed due to unavailable pixel data. Verdict: strong suspicion of data fabrication and methodological misrepresentation; formal investigation recommended.

Verdict

🔴 Strong suspicion of data fabrication and methodological misrepresentation. Multiple independent anomalies in tabular and methodological content are mutually reinforcing. Formal institutional and editorial investigation is warranted.

Key findings

Implausible ZDOCK score for ErbB1 (13553.6) exceeding the algorithmic range of rigid-body FFT docking, while all other receptors fall within 1528–1967.
Identical binding energy of −17.7 kcal/mol reported for both ErbB1 and ErbB2 in Table 1, indicative of copy-paste duplication.
Methods section (5.8) cites a PerkinElmer PinAAcle 900T for ICP-OES measurements; PinAAcle 900T is an atomic absorption spectrometer (AAS), not an ICP-OES instrument.
STRING protein-interaction network p-value of 0.0173 for a 37-node (Src + 36 partners) network is implausibly high relative to typical STRING enrichment outputs.
Image reuse / splicing analysis (Findings 1 and 3 of detection suite) could not be performed due to absence of pixel-level image data.

Evidence highlights

Table 1 / Page 9 — ZDOCK scores: ErbB1 = 13553.6; BMPR1A = 1528.46; BMPR1B = 1967.40; BMPR2 = 1601.76; ErbB2 = 1937.10; ErbB3 = 1778.69; ErbB4 = 1879.33. The single outlier is ~7–9× the magnitude of peers.
Table 1 / Page 9 — Binding energies: ErbB1 = −17.7 kcal/mol; ErbB2 = −17.7 kcal/mol (exact match to one decimal place).
Methods 5.8 / Page 15 — Stated technique: ICP-OES. Stated instrument: "PerkinElmer PinAAcle 900T" (an AAS platform).
Page 6 (text) — STRING v11 network: Src as central hub interacting with 36 partners; reported p = 0.0173.
DOI: 10.1002/adfm.76607

Notes

The ZDOCK outlier alone is sufficient to cast serious doubt on the molecular-docking component; authors should be asked to provide raw ZDOCK output logs.
The ICP-OES / AAS mismatch is either a critical methodological error or evidence that the described measurement was not actually performed with the cited instrument; raw instrument printouts should be requested.
The p = 0.0173 anomaly is suggestive but depends on STRING algorithm version specifics; should be re-verified by re-running STRING v11 with the same gene list.
Image-level verification (Western blots, immunofluorescence, histology) is strongly recommended once raw TIFF files are obtained, given the pattern of tabular irregularities.
All findings derive from textual and tabular content of the published PDF; pixel-level image forensics were not conducted.