Geng integrity report — 'A lncRNA fine-tunes salicylic acid biosynthesis to balance plant immunity and growth' (Liu et al., Cell Host & Microbe, 2022)

Academic fraud report · Geng Detector

Summary

Verdict: Questionable (🟡). The report flags substantive methodological and statistical concerns in this 2022 Cell Host & Microbe paper on SABC1 lncRNA. The most serious issue is the misuse of statistics in Figures 4 and 6: Figure 4I is explicitly described in the legend as having error bars representing SD from only 2 technical replicates, yet Student’s t-tests with significance stars are reported. A t-test at N=2 is mathematically unreliable. Figure 6D cites 15 technical replicates for a dual-luciferase assay, again substituting technical precision for biological replication. Secondary concerns include ambiguous fold-change phrasing ('0.7-fold higher', '0.8- to 1.3-fold increase') in Results, which is non-standard and potentially misleading. Two further claims (a statistical-assumption disclaimer and image duplication) could not be corroborated from the available text and remain unverified. Image-level forensic analysis was not possible because only text was supplied. The flagged issues are real and reproducible from the PDF but do not, by themselves, prove fabrication.

Verdict

Questionable (🟡). The paper contains clear and verifiable misuse of statistical replicates in key mechanistic figures (Figure 4, Figure 6) and ambiguous quantitative wording in Results. These issues, taken together, justify public scrutiny and a request for raw biological replicate data. No visual image-manipulation analysis could be performed because high-resolution figures were not available.

Key findings

Figure 4I uses N=2 technical replicates for a t-test. Legend states "Error bars represent SD from 2 (I) or 3 technical replicates (E–H, J, and K)." Significance stars derived from Student’s t-test at N=2 are statistically unreliable (very low degrees of freedom, prone to false positives/negatives).
Figure 6D dual-luciferase assay uses 15 technical replicates ("Error bars represent SD from 15 technical replicates"). Technical replicates reflect instrument precision, not biological variance; statistical inference from them alone is inappropriate for the mechanistic claims made.
Ambiguous fold-change language in Results: "SA accumulation in sabc1-1 and sabc1-2 was 0.7- and 1.0-fold higher than that in the Col-0" and "the transcript level of NAC3 showed a 0.8- to 1.3-fold increase." Standard convention favours "increased by X%" or "X-fold of." Repeated non-standard phrasing raises concerns about data handling transparency.
Unverified claims (labelled ⚠️ in the original report): (a) the cited statistical-assumption disclaimer in STAR Methods — the text provided contains only the section heading "QUANTIFICATION AND STATISTICAL ANALYSIS" and "ImageJ image analysis"; the quoted sentence could not be located. (b) image duplication/pixel-level manipulation — not assessable from text alone.

Evidence highlights

DOI: 10.1016/j.chom.2022.07.001 (Cell Host & Microbe, 2022).
Figure 4 legend (verbatim): "Error bars represent SD from 2 (I) or 3 technical replicates (E–H, J, and K)." Panel I involves ChIP-qPCR supporting CLF binding and H3K27me3 modification — a central mechanistic claim.
Figure 6D legend (verbatim): "Error bars represent SD from 15 technical replicates" — applied to dual-luciferase reporter assay.
Results text: "SA accumulation in sabc1-1 and sabc1-2 was 0.7- and 1.0-fold higher than that in the Col-0"; "the transcript level of NAC3 showed a 0.8- to 1.3-fold increase."
STAR Methods (provided excerpt): Contains "QUANTIFICATION AND STATISTICAL ANALYSIS" and "ImageJ image analysis"; the alleged disclaimer sentence was not located in the supplied text.

Notes

Confidence: High for findings 1 and 2 (text-based, reproducible). Low for findings 3 and 4 (text excerpt insufficient).
Limits: No image-level (Western blot background, splicing, microscopy noise) analysis was conducted because only the PDF text/legends were available.
Recommended follow-up: (i) Request raw biological replicate data (independent experiments, N ≥ 3) for ChIP-qPCR (Fig. 4) and dual-luciferase assays (Fig. 6D) from the corresponding authors; (ii) File a PubPeer comment focused on the misuse of technical replicates for inferential statistics; (iii) Consider notifying the journal's editorial integrity office about the statistical-methods concern. This report is informational and does not constitute a formal finding of misconduct.