Investigation Report: Recyclable amyloid-based magnetic nanonets for active capture and removal of nanoplastics from water (DOI: 10.1038/s44221-026-00620-1)

Academic fraud report · Geng Detector

Summary

Verdict: QUESTIONABLE (not a confirmed misconduct finding). This Nature Water paper covers an unusually broad scope—synthesis, multi-technique characterization, DFT/MD simulations, in vivo mouse imaging, and life-cycle assessment (LCA)—which itself raises plausibility questions regarding author attribution. Key concerns include: (1) implausibly low standard deviations in co-pollutant adsorption data reported as n=3 independent replicates (e.g., 38.6 ± 0.02 mg g⁻¹), inconsistent with the biological/environmental variability expected in bacterial and natural-water systems; (2) a clear typographical error in which the abbreviation LAF-IONPs is rendered as 'LAF-INOPs' in the ICP-OES description, suggesting copy-paste from another draft; (3) the author-contribution statement claims a single co-author performed 'all experiments,' spanning XAFS analysis, GROMACS MD, DFT, IVIS imaging, and LCA—specializations typically requiring distinct expertise; (4) animal experiments used n=3 mice per group, the minimum statistical threshold, yielding an extremely low p-value (5.05 × 10⁻⁵). Limitations: this analysis is based solely on the provided PDF/OCR text; no raw image files were available, so image-duplication and pixel-level forensics were not performed. Findings are suggestive but not conclusive.

Verdict

🟡 Questionable. Several methodological and reporting inconsistencies warrant author clarification, but the current textual evidence is insufficient to confirm data fabrication or misconduct.

Key findings

Implausibly low standard deviations: Co-pollutant adsorption capacities (Figure 6a) are reported with SDs as small as ±0.02 mg g⁻¹ (e.g., control: 38.6 ± 0.02) and ±0.1 mg g⁻¹ (SiO₂: 40.0; methyl orange: 40.2) for what are described as n=3 independent replicates involving bacterial and natural-water matrices. Such precision is biologically unrealistic unless the replicates are technical (e.g., same-solution absorbance readings) rather than biological repeats.
Typographical/copy-paste error: In the Characterization section (Page 3, right column), the established abbreviation LAF-IONPs is rendered as LAF-INOPs, indicating likely copy-paste from an unrelated draft.
Implausible single-author scope: The Author Contributions section credits Xinyu Yu with performing 'all experiments,' yet the manuscript spans XAFS analysis, GROMACS-based MD simulations, DFT calculations, in vivo IVIS imaging, and life-cycle assessment—competencies that are normally distributed across multiple specialists.
Marginal animal-group sample size: Mice were divided into three groups with n=3 per group, the statistical minimum. The reported p-value of 5.05 × 10⁻⁵ (Figure 6f) is unusually small for such low n, raising questions about variance and effect-size robustness.

Evidence highlights

DOI: 10.1038/s44221-026-00620-1
Quoted values (Figure 6a, Page 7):
SiO₂ group: 40.0 ± 0.1 mg g⁻¹
Methyl orange group: 40.2 ± 0.1 mg g⁻¹
Control: 38.6 ± 0.02 mg g⁻¹
Typo evidence: LAF-INOPs instead of LAF-IONPs (Page 3, right column, ICP-OES description).
Author contribution claim: X.Y. performed all experiments. (Page 12).
Animal protocol: Mice were randomly divided into three groups (n = 3 per group) (Page 10).
Statistical outcome: P = 5.05 × 10⁻⁵ (Figure 6f).

Notes

Pixel-level image-duplication and Photoshop-tampering checks (the 'first' and 'third' forensic techniques) could not be executed because no raw image files were supplied; only PDF/OCR text was analyzed. Absence of alerts here does not imply image integrity.
The combination of suspiciously precise error bars with a low-effort typo is a recognized red-flag pattern, but it is also consistent with careless drafting on an otherwise legitimate dataset. Independent verification of the underlying raw fluorescence/absorbance values is the decisive next step.
Reporting n=3 per group is permissible under most journal policies, including Nature Water's, but is below the conventional robustness expected for in vivo biodistribution claims in a high-impact venue.
Recommended actions: request raw numerical data and replicate-level metadata from the corresponding author; consider a PubPeer comment focused on clarifying whether Figure 6a replicates are biological or technical. Institutional or editorial escalation is not warranted on present evidence.