After correcting a paper due to problematic figure panels, researchers led by high-profile biologist Olivier Voinnet have now retracted it, after “further analysis of the paper revealed flaws in the interpretation of” another figure.
PLOS Genetics published the retraction notice September 3 for the 2013 paper on the molecular details of embryonic stem cells in mice. First author Constance Ciaudo and Voinnet assume “full responsibility for the mistakes on this paper,” according to the note.
Questions about the figures in “RNAi-Dependent and Independent Control of LINE1 Accumulation and Mobility in Mouse Embryonic Stem Cells” had been raised on PubPeer earlier this year; the paper was among many other papers co-authored by Voinnet that were questioned on the site.
Shortly after the first PubPeer post about the 2013 PLOS Genetics paper, a commenter using the name of first author Ciaudo said she had become aware of “potential problems” with a figure, and that Voinnet was contacting the journal’s editor.
One month after issuing the correction, PLOS Genetics posted an Expression of Concern about the paper, which mentions the same figure panels included in the correction, along with another figure:
Concerns have been raised regarding some of the data in the PLOS Genetics article “RNAi-Dependent and Independent Control of LINE1 Accumulation and Mobility in Mouse Embryonic Stem Cells”, notably in panel A in Fig 4, in panels A and F in S4 Fig, and with the statistical analyses used to produce Fig 2. The authors have responded to these concerns, acknowledging that some errors were made in figure preparation and with some statistical tests, however a final resolution has not yet been reached, and the matter is being evaluated by the authors’ institution.
Now, a couple of months later, the paper has been retracted. Here’s the full retraction notice:
At the request of the authors, PLOS Genetics is retracting this publication following an investigation into concerns that were raised regarding the assembly of Fig 4 and S4 Fig, and the statistical analysis used in Fig 2A. The text below has been agreed to by the authors and editors.
The corresponding author, Olivier Voinnet, was originally alerted to errors that occurred during the assembly of Fig 4 (panel A) and S4 Fig (panels A and F). These errors have been corrected using the original raw data, and a correction notice was published accordingly.
Further analysis of the paper revealed flaws in the interpretation of the transposition data presented in Fig 2A. In the originally submitted version, the L1 copy number was only presented for the DCRFlx/Flx P10 and DCR-/- P30 cells, and a T-test performed on the two datasets showed that the L1 copy number was statistically higher in DCR-/- cells than in control cells. During the last stage of the review process, additional datasets were added and a second T-test was then used to establish the statistical analysis published in the final version of the paper. However, it was later realized that T-tests are not appropriate for comparing more than two datasets. At the recommendation of the ETH statistics helpdesk, a suitable Analysis of Variance (ANOVA) test with multiple comparisons was then conducted on the DcrFlx/Flx P30 and Dcr-/- P30 datasets, providing a p-value of 0.0501, which is at the margin of the threshold of significance. The ANOVA test conducted on the DcrFlx/Flx P10 and Dcr-/- P30 datasets revealed a statistically significant p-value of 0.0018. The statistical issue regarding the L1 copy number in DCR-/- versus control ES cells is currently being addressed using a new set of cells and a direct GFP-based transposition assay. This issue will hopefully be clarified in the near future via the submission of an amended study for peer-review.
Based on the present uncertainty revealed by the corrected statistical analysis of the L1 copy number—a key element of this paper—and on the previous errors in the figures, the authors have collectively decided to retract this study. Constance Ciaudo and Olivier Voinnet take full responsibility for the mistakes on this paper and wish to apologize. They also wish to state that none of the above-mentioned mistakes involved any of the co-authors from the Curie Institute, whose contributions to the paper were restricted to the bioinformatics analysis of small RNAs (NS, CJC, EB) and the generation of reagents including an ES cell line required for the study (EH, IO). All authors regret the inconvenience caused.
The study has been cited 16 times, according to Thomson Scientific’s Web of Knowledge.
Our total count for Voinnet now stands at 12 corrections (including one from this newly retracted paper) and five retractions.
In July, Voinnet was suspended from Centre national de la recherche scientifique (CNRS) for two years following investigations into problems with his research. An inquiry by ETH Zurich found “errors of varying severity” in 20 of the 32 studies they looked into.
We’ve reached out to Voinnet and Ciaudo for a comment on the retraction. We’ll update with any response.
Update, 10:00 a.m. Eastern, 9/8/2015: In a statement on the retraction, a spokesperson for PLOS Genetics said that they were tipped off by a “member of the community”:
The overall history of this retraction is given in the retraction notice itself, and we hope that readers will find this to be a clear and informative explanation. Prompted by an alert from a member of the community, and after further investigation by the authors and journal editors (during which time an Expression of Concern was posted to inform readers), it was felt that the conclusions of this article were no longer considered reliable. Following this decision, PLOS Genetics supported the authors in retracting the article, which we believe was the responsible step to take.
As a community journal, we rely on and value the input of our readers, editors and authors. If readers have concerns about research published in PLOS Genetics, we would always encourage them to contact the journal office directly. We investigate all concerns in accordance with the guidelines issued by the Committee on Publications Ethics (COPE) and aim to resolve questions promptly.
Like Retraction Watch? Consider making a tax-deductible contribution to support our growth. You can also follow us on Twitter, like us on Facebook, add us to your RSS reader, and sign up on our homepage for an email every time there’s a new post. Click here to review our Comments Policy.