Geng Academic Integrity Review: "A Functional Switch of NuRD Chromatin Remodeling Complex Subunits Regulates Mouse Cortical Development" (Cell Reports, 2016)

Academic fraud report · Geng Detector

Summary

Verdict: ✅ Likely clean (text-based preliminary analysis). This review of Nitarska et al., Cell Reports 2016 (DOI: 10.1016/j.celrep.2016.10.022), found no textual indicators of academic fraud. The authors transparently disclosed shared control data across Figures 5 and 6 in the legend—an unusual but exemplary practice that preemptively removes suspicion of covert image reuse. Methodological details (Q-Exactive mass spectrometer, UniProt SwissProt May 2010 database) are temporally consistent with a 2016 submission. Reported transcript counts (3,627 transcripts / 2,835 DE genes) are biologically plausible (multiple transcripts per gene). Statistical methodology (t-tests, one- and two-way ANOVA with Tukey/Sidak post-hoc) is appropriate for the experimental designs described, with sample sizes (n=3–15 embryos) typical for developmental neurobiology. Confidence: low-to-moderate due to text-only analysis. Limitations: pixel-level image forensics, original Western blot inspection, and verification of the GEO microarray dataset (GSE70298) were not performed.

Verdict

✅ Likely clean based on text-level analysis. No indicators of academic misconduct were identified. The most notable finding is that the authors proactively disclosed reuse of control data between Figures 5 and 6, which is best practice and removes any suspicion of concealed duplication.

Key findings

Transparent disclosure of shared data (Figure 6 legend): The authors explicitly state that "shCTL, shCHD5, and shCHD5 + hCHD5 conditions are identical to those shown in Figures 5A and 5B, because the experiments were performed at the same time, and data were split in two figures for clarity." Similar statements appear for Figures 6C/D and 6G/H.
Temporally consistent methodology: Use of the Q-Exactive mass spectrometer (released 2011–2012) is appropriate for a 2016 publication; the UniProt SwissProt May 2010 database is slightly dated but acceptable for reproducibility.
Biologically plausible expression data: 3,627 differentially expressed transcripts corresponding to 2,835 genes is consistent with alternative splicing producing multiple transcripts per gene.
Appropriate statistical design: Combination of unpaired t-tests, one-way ANOVA, and two-way ANOVA with Tukey or Sidak post-hoc multiple-comparison corrections is appropriate. Sample sizes (n = 3–15 embryos) are realistic for the experimental system.

Evidence highlights

DOI: 10.1016/j.celrep.2016.10.022
Figures referenced in transparency statements: Figures 5A, 5B, 6A, 6C/D, 6G/H (all sharing shCTL / shCHD5 / shCHD5 + hCHD5 conditions)
Microarray transcript counts: 30,869 total transcripts, 3,627 differentially expressed, corresponding to 2,835 genes
Mass spectrometry: Q-Exactive instrument, ProLuCID algorithm, UniProt SwissProt (May 2010 release)
Suggested raw data repository for follow-up: GEO accession GSE70298

Notes

This assessment is text-only; pixel-level image analysis (Western blot edges, immunofluorescence stitching) and numerical statistical forensics (Benford's law, last-digit distribution) were not possible.
The transparent flagging of shared control panels, rather than being a red flag, is in fact a marker of careful authorship and editorial diligence.
Recommended follow-up (not performed here):
Obtain a high-resolution PDF and inspect Figures 1 and S1 Western blot bands for sharpening artifacts.
Download the GEO GSE70298 expression matrix and verify variance distribution and DE gene counts.
AI-assisted; for any formal allegation, an institutional investigation with full original data and images would be required.