The “Voinnet Case”: Incorrect Measures for Incorrect Figures?

(July 12th, 2015) Formal investigations by the ETH Zürich and the French CNRS identified numerous incorrect figures in about twenty publications by plant biologist Olivier Voinnet. Nevertheless, he is to keep his laboratory and professorship at the ETH. Hard to understand, as many colleagues express their astonishment.

The ETH Zürich and the French CNRS have concluded their formal investigations into the accusations of scientific misconduct against the French plant biologist, Olivier Voinnet. Both institutions, where Voinnet currently runs active laboratories, have presented the results and their decisions of disciplinary actions in two separate press releases (here and here). The ETH has also released the entire report of its Commission of Inquiry.

Here are the results in a nutshell:

- ETH and CNRS found numerous incorrect figures in about 20 of Voinnet’s publications clearly breaching the ethical standards of publication.

- Altogether, the inappropriate presentation of experimental data, however, does not amount to fabrication and, at least according to the ETH, does not represent a case of scientific misconduct.

- The scientific conclusions of those papers are not being questioned. Yet, the ETH report recommends the retraction of five articles and corrections of 15 others.

- The CNRS intends disciplinary measures, the first one being that Voinnet has been excluded for two years from the CNRS.

- Voinnet is to keep his lab and professorship with the ETH but “is to be provided with an external advisor for the necessary changes in working practices”.

What to make of it all?

In April, at the beginning of the investigation, the ETH announced that only Voinnet’s “illustrations” might be problematic, while his findings were not in doubt. Apparently capable of seeing into the future, this was exactly what ETH has now “found”. Their conclusion: what Voinnet was definitely guilty of actually ranged from "the ’beautification’ of figures to the mere confusion of correct and incorrect figures”; some figures, for example, “had been edited for internal use and then published in error”. The ETH commission further stated that this “processing of the figures led to no scientific advantage” and that the “raw data and documentation from the experiments […] were complete and correct and substantiated the scientific conclusions made in the publications”. Therefore, “the Executive Board of ETH Zurich has concluded that this is not a case of scientific misconduct as defined by the Rules of Procedure”.

This statement of the Executive Board appears to be in contrast to the report of its own Commission of Inquiry. First of all, unlike the press release suggests, the commission investigated 32 Voinnet publications from 1998 to 2013, not just those five that originated from his lab at ETH. But most importantly, the commission clearly states in its report that Voinnet was indeed found guilty of misconduct (“The majority of problematic papers contained clearly identified cases of misconduct”, p. 17). He himself, however, blamed it on “exciting but high-pressure environment at the forefront of science”, leading to his papers being “assembled too quickly, with 'no moment of reflection', in a highly competitive environment”.

The scientific investigation was performed by Matthias Peter and Yves Barral from the ETH’s Department of Biology as well as two external experts, Edward Farmer (University of Lausanne) and Witold Filipowicz (Friedrich Miescher Institute, Basel). This committee investigated the accusations against Voinnet along four categories of “research malpractice”. In the accused papers it found no evidence of malpractice of Category 1, namely “the invention of data in the place of experiments that were never conducted or manipulation of data with a purpose to change the conclusions”. The argument went: Voinnet was apparently always able to present the corresponding original data. In fact, the report repeatedly stated: “It was not clear to the commission why these manipulations were done since inspected primary data looked similar and introduced changes generally had no effect on conclusions of the experiment.”

Categories 3 and 4 were reserved for sloppiness, such as undisclosed image splicing (which according to the report was acceptable until the years 2006/2008) and unintended publication of erroneous images in place of the correct ones. Into the latter category apparently also went the so-called “’mock’ idealized figures”. According to Voinnet those “were not destined for publication but represented provisional figures used by lab members during internal lab meetings with the aim to facilitate presentation or sketching figures for prepared manuscripts”. These figures “were then 'mistakenly' used in the publication”.

To the commission, this bizarre explanation obviously sounded reasonable enough, even though, among other things, it takes much more time, skill and effort to convincingly splice and manipulate a “mock” figure in Photoshop, than to assemble a correct one from unmanipulated images. The investigators, however, did repeatedly reprove Voinnet in the report for this “bad habit of using false idealized figures instead of primary raw data during lab discussions”.

Finally, the commission found several cases of Category 2 misconduct, meaning “deliberate modification, duplication, or mislabeling of images in order to make them look nicer or more convincing”. Despite insisting that these manipulations occurred “without however affecting the overall conclusion of the original experiment”, the commission considered it as necessary to wipe out the entire scientific content of five “Category 2” publications by demanding their retraction. Those five are: Sansregret et al., PLoS Pathogens, 9(6): e1003435; Deleris et al., Science 313: 68-71; Moissiard et al., PNAS 103(51): 19593-8; Dunoyer et al., EMBO J 29(10): 1699-712; and Dunoyer et al., Plant Cell 16(5): 1235-50, the latter of which was actually retracted in June.

For three more papers of that “Category 2”, the commission explicitly does not propose retractions, but corrections, “given that primary data exist and experiments were correctly conducted and documented”. These papers, which were found to share one and the same loading controls in different contexts, are Schott et al., EMBO J 31(11): 2553-65; Dunoyer et al., Nature Genetics 39(7): 848-56 and Dunoyer et al., Science 328: 912-6. Ironically, shortly after the committee made this judgement, the latter publication was again flagged on PubPeer with the additional finding of duplicated bands. It is not the only case where one is left to wonder if, in fact, a retraction might be the more appropriate option.

The ETH, nevertheless, appears to be determined to support Voinnet. In his apology for “not exercising the necessary care during the publication process“, Voinnet also thanks the university “for supporting me despite these errors” and for having “placed its trust in me”. Voinnet’s lab in Zürich with its 30 employees is now instructed to carry “electronic laboratory notebooks” and will be supervised by “an external advisor”. All their future publications are to be monitored and reviewed by the Biology Department Executive Committee and the Executive Board of the ETH.

In contrast to the measures of the ETH, the reaction from the French seems much harsher. The CNRS press release stated that Voinnet’s “deliberate chart/diagram manipulations” do amount to a “breach of the ethical standards” and “scientific misconduct”. Because these violations “have tarnished the reputation of the CNRS and of research at large”, disciplinary measures are to be taken. The most prominent one is already listed in the investigative report: Voinnet is to lose his laboratory at the CNRS Strasbourg which has ten employees and is currently run on his behalf by Patrice Dunoyer, co-author on numerous of the evidently problematic papers.

According to a report by the French newspaper Liberation, the CNRS investigation has “found manipulation of figures in thirteen articles”, for which Voinnet assumed responsibility. The prizewinning French scientist, who is currently on secondment from CNRS to the ETH, is now "excluded for a period of two years from the decision to terminate his secondment" – meaning he cannot return to CNRS anytime soon. In fact, the article suggests that Voinnet’s ban from French research is thus likely to be permanent.

On an aside, however, a deeper look into the report by the ETH commission also reveals some further peculiar nuggets, which are mostly about Voinnet pretending to be a victim of natural disasters - of others’ misconduct and of a general conspiracy against himself.

For example, Voinnet rebutted criticisms on immunoblot lane duplications in the Marí-Ordóñez et al. paper (Nature Genetics, 45: 1029-39) with the argument that “although original phosphorimages were lost when a room at the Strasbourg laboratory was flooded, the original membranes used to make Figs. 5F and 6A were found and re-probed”. The commission eagerly took Voinnet’s word that the new images do match the ones swallowed by the floods.

With regard to the Sansregret et al. paper (PLoS Pathogens, 9(6): e1003435), Voinnet actually “admitted that Figure 6 was fabricated and that this was for him the 'worst case'”. Yet he insists that “a Master's student” fabricated the panels in question, which neither Voinnet nor other authors noticed when assembling a revised manuscript version. The first author, Raphaël Sansregret, who at the time was a PhD student in the Canadian lab primarily responsible for the paper, had told Lab Times already in January 2015 that the figure in question was definitely generated in the Voinnet lab. Accordingly, Dunoyer also stated in January on PubPeer that it was he who “used a figure that was not meant to be published”.

When defending the retraction-bound Deleris et al. paper (Science 313: 68-71), Voinnet provided several alternative explanations, including cheating students. The report reads “that this paper did not contain 'mock' idealised figures, as originally believed by the commission, but mistakes which occurred during mounting of composite figures from primary data contributed by two different graduate students”. In the end, Voinnet admitted that it was actually he who was solely responsible for all errors in this paper.

Most strikingly, Voinnet was apparently asked by the investigators if there had been “an organised effort to discredit OV and inflate accusations against him”. And indeed, Voinnet even presented a named culprit. According to the investigative report, Voinnet said that “although he has no proof of it, he has the strong suspicion that a colleague with whom he had a scientific conflict at Cell Reports (9: 795 and 9: 798) might have helped orchestrate the sudden dissection of all his scientific production”. The second publication (Vol. 9(3):798-99) is a criticism of the first (Ding and Voinnet, Vol. 9(3):795-97) and was written by only one author: Benjamin tenOever, Professor of Microbiology at Mount Sinai Hospital, New York.

In response to Voinnet’s allegation, tenOever immediately sent the followings statement to Lab Times:

It has come to my attention that the ETH has published an official report investigating issues of scientific misconduct derived from the laboratory of Olivier Voinnet. For reasons I do not understand, I am accused in this document of being someone who may have “helped orchestrate the sudden dissection of all his scientific production”. This statement is completely false. I have not produced any written criticism of the work of Olivier Voinnet outside of the academic arena of my peer-reviewed publications and their associated press releases. Furthermore, any discussions of our scientific disagreement have been made in the spirit of academic discourse and have been confined to professional settings which most certainly do not constitute grounds for such a proposed grand conspiracy. I am very disappointed that the ETH permitted the inclusion of such an unfounded statement. Such accusations border on libel and only serve to distract from the actual issue of scientific misconduct.”

This, however, was by far not the only immediate reaction on the issue from the scientific community. Predominantly on PubPeer and Twitter, many plant biologists instantly expressed a strong discontent with the conclusions of the investigations and the measures now taken against Voinnet by ETH and CNRS. Almost all of them make no secret of the fact that in their opinion Voinnet seems to be getting off the affair far too lightly. An anonymous commenter, for example, stated on PubPeer, “A very inadequate verdict and a devastating message to all ‘real’ young (as well as old) scientists out there.” David Logan, plant biology professor at the French University of Angers, commented on Twitter, “Goes against everything I've been taught about performing experiments, mentoring students and publishing results.” And Sophien Kamoun from the Sainsbury Lab in Norwich added, “One could say the idiots are the ones who followed the rules and didn't get the papers, jobs and grants.”

Some others even turned to cynicism. Ralf Reski, professor for plant biotechnology at the University of Freiburg, asked via Twitter, “So, is Olivier Voinnet now the Lance Armstrong of Plant Science?” And it was again Sophien Kamoun who succinctly summed up the results of the investigations as a “sad day for plant biology”.

So, regardless of whatever official measures are now to be taken, it seems that Voinnet’s reputation in the plant science community has already plummeted to zero, anyway.

Leonid Schneider


Last Changes: 08.25.2015

Information 4

Information 5