Geng Academic Fraud Review: "Long non-coding RNAs are involved in alternative splicing and promote cancer progression" (British Journal of Cancer, 2022)

Academic fraud report · Geng Detector

Summary

Verdict: CLEAN (no fraud indicators detected). This paper, DOI 10.1038/s41416-021-01600-w, is a Review Article with no author-generated primary experimental data. Figures 1–4 are schematic mechanism illustrations drawn by the authors, and Table 1 is a literature summary table. Therefore, the original-data-dependent Geng checks (image reuse, numeric fabrication, image splicing, statistical anomalies) are largely inapplicable. Machine-forensic copy-move signals in images I2/I4/I6/I8 correspond to regular schematic structures (exon boxes, intron arcs, splice-factor icons, panel borders) rather than experimental image duplication. Productivity/timeline and citation/methodology consistency checks both passed. A mild AI-polishing signature (score 0.20/1.0) was detected in the text, consistent with non-native English grammar editing and not indicative of misconduct. Confidence is high for the applicable checks; limitations arise from the fact that several checks cannot be applied to a review article.

Verdict

CLEAN. No academic fraud indicators were identified in this Review Article. The article contains no original experimental data, which structurally limits the applicability of data-dependent fraud checks. All applicable checks passed.

Key findings

No original data present: The article is a Review Article. Data Availability statement reads "Not applicable"; no Western blots, microscopy images, or original statistical charts appear across the 12 reviewed pages.
Figures 1–4 are author-drawn schematics; Table 1 is a citation summary. The four copy-move forensic signals (I2, I4, I6, I8) are explained by regular schematic motifs (exon boxes, intron arcs, splice-factor symbols, panel frames), not image tampering.
Noise-variance asymmetries (I1, I3, I5, I7) reflect intrinsic grayscale/texture distribution of schematics; no homogeneous baseline exists for review figures, so they do not constitute anomalies.
Productivity/timeline check passed: Submission 2020-12-06, accepted 2021-10-11, published 2021-11-08 — a reasonable review timeline. Funding IDs (NSFC 82073135, 82072374; Hunan 2019JJ50354, 2019JJ50780; CSU 2021zzts0922) are normally formatted.
Citation/internal-consistency check passed: All lncRNA–AS cases in the text cross-reference correctly with Table 1 and the reference list (e.g., PNUTS → [29], PD-L1-lnc → [30], ORAOV1-B → [31], PXN-AS1 → [36], PVT1 → [37]). Citation span 2008 (Beltran M, ZEB2-anti [44]) to 2021 (Pruszko M, VEGFA [58]; Bast-Habersbrunner, Ctcflos [85]) is consistent with a review.
Mild AI-polishing signature: Text metric T1 score = 0.20/1.0 (light grade), all other text metrics clean. Consistent with grammar/AI-assisted polishing common in non-native English academic writing; not evidence of fraud.
Programmatic cross-check: 3 citation data points tested, 3/3 matched source text (per the original report's verification log).

Evidence highlights

DOI: 10.1038/s41416-021-01600-w — verified.
Authors: Jiawei Ouyang, Yu Zhong, Yijie Zhang, Liting Yang, Pan Wu, Xiangchan Hou, Fang Xiong, Xiayu Li, Shanshan Zhang, Zhaojian Gong, Yi He, Yanyan Tang, Wenling Zhang, Bo Xiang, Ming Zhou, Jian Ma, Yong Li, Guiyuan Li, Zhaoyang Zeng, Can Guo, Wei Xiong (corresponding: Can Guo, Wei Xiong, Central South University).
Author contribution statement: ten authors listed under "collected the related paper and finished the manuscript and figures" — consistent with review writing workflow.
Forensic image flags (I2, I4, I6, I8): all judged benign based on schematic content.
Text metric: T1 = 0.20 (light AI-polishing signal only).

Notes

This assessment is limited by the article type: several Geng fraud checks (image reuse, numeric fabrication, image splicing, statistical anomalies) cannot be meaningfully applied to a review without primary data.
The review's intellectual value — quality of selection and synthesis of lncRNA–alternative splicing literature — lies outside the scope of fraud detection and should be judged by domain peers.
The mild AI-polishing signature is documented as an observation, not as a fraud indicator.
All findings are derived from the supplied Chinese report; no independent image re-analysis was performed.