A Caltech researcher who shared the 2018 Nobel Prize in Chemistry has retracted a 2019 paper after being unable to replicate the results.
Frances Arnold, who won half of the 2018 prize for her work on the evolution of enzymes, tweeted the news earlier today:
The paper has been cited once, according to Clarivate Analytics’ Web of Science.
The notice went live just a few hours after Arnold’s tweet:
After publication of the Report “Site-selective enzymatic C‒H amidation for synthesis of diverse lactams” (1), efforts to reproduce the work showed that the enzymes do not catalyze the reactions with the activities and selectivities claimed. Careful examination of the first author’s lab notebook then revealed missing contemporaneous entries and raw data for key experiments. The authors are therefore retracting the paper.
The first author, Inha Cho, is no longer working in the lab, Arnold told Retraction Watch.
‘Very appreciative’ of ‘forthcoming and proactive’ approach
Holden Thorp, the editor in chief of Science, said that Arnold contacted the journal on Dec. 1 and the Science had planned to publish the retraction next week, along with the printed version in the Jan. 10 issue.
Dr. Arnold tweeted it today so we went ahead and put it up.
He added:
As I said on Twitter, we are very appreciative of Dr. Arnold and coworkers and the forthcoming and proactive way they have dealt with this.
Arnold told Retraction Watch that her lab was
not able to reproduce the results described in the paper. As we looked deeper we could not find contemporaneous data in one of the coauthor’s notebooks to support the claims.
She added:
All three coauthors agreed to the retraction and have apologized.
I was in the middle of all the Nobel Prize hoopla and did not pay enough attention to this submission, so it is my fault.
She will hardly be the first Nobelist to retract a paper. The list includes Linda Buck (whose co-author was sanctioned by the U.S. ORI several years later), Michael Rosbash, Jack Szostak, and Bruce Beutler. Daniel Kahnemann has also publicly admitted errors, although he has not retracted any scholarly papers. And Paul Nurse issued a correction a few years ago.
As we noted in a STAT column last year:
The willingness to admit mistakes publicly sends a message of trustworthiness to fellow researchers.
Like Retraction Watch? You can make a tax-deductible contribution to support our work, follow us on Twitter, like us on Facebook, add us to your RSS reader, or subscribe to our daily digest. If you find a retraction that’s not in our database, you can let us know here. For comments or feedback, email us at [email protected].
http://fhalab.caltech.edu/?page_id=392
Professor Arnold has 14 graduate students, 10 post-docs, and 1 staffer (and probably any number of undergraduate interns). How can anyone possibly manage 24 researchers?
Agreed with Andy Patterson. Dissipation of project supervision is a real problem.
Some groups have over 50 researchers…
I think that 15 is the optimal group size.
Really? Most faculty cannot properly manage one—to make this individual very productive—in my experience. Maybe you need 15 (30 hand with pipettes) in a lab to make up for the failure of the faculty member to properly manage and make an individual reasonably productive.
I agree with some of the posters above that this lady is getting too much credit for doing the bare minimum by retracting this paper.
(a) She should frankly be ashamed of herself for having a group that is so large that she has no time to manage / mentor them properly. I wonder why the NSF / NIH do not penalize excessively large groups when considering grant awards? It is eminently doable, and probably the only way to check the devolution of scientists into glorified lab managers / used-car salesmen.
(b) In cases like this (missing / potentially fabricated data), a FULL disclosure is required, not just an “Oops, I’m sorry…”. The first author of this paper, in particular, should be investigated.
(c) The referees of the retracted paper and the Science editor who handled it need to lose anonymity: They messed up. Seeing a “big name” corresponding author is no excuse to fail due diligence. Measures like this are essential to smash the cozy and incestuous clubs that pretend to conduct peer-review in these fancy venues.
If you don’t have a Manhattan-project sized scientific empire, the US funding agencies don’t think your work is important or fundable. The days of “thinking scientists” are long gone. If you don’t have mega-bucks to throw at a problem and publish a papers with 60 panel figure and 200 co-authors, you are insignificant.
One need not be a Nobel laureate nor a full professor to be a good team leader and supervisor and I expect Prof. Arnold to have several of those, at a rough guess three or four, to help her manage her team. I couldn’t do it, which is why I will never rise to any position of leadership, but it is doable and I have seen it being done.
Is it so easy to give a nod to junk science with the excuse: ‘oh, I’m so busy… yeah, let the article fly’? So: junk science gets published and you don’t care about the science because you’ve become a big honcho.
Really.
I think she realizes letting it fly was a mistake, and she showed she cares by retracting the article.
It is a very messy situation, these so called top scientists get large sums of grant money, hire scores of scientists who compete with each other to please the boss who has no time to look at the data or the written manuscripts because majority of their time is wasted on flying and delivering the same lecture at different institutes. Let us support more young scientists with better ideas and aptitude to achieve more than these so called established scientists.
Young scientists like Inha Cho?
It seems to me that, in the rush to defend Prof. Arnold and praise her for her transparency in admitting her mistake, she and her lab might be getting off too easy. At first, I was fine with the explanation that they had just gotten it wrong. Everyone, myself included, makes mistakes, and occasionally big ones. However, then I read that “Careful examination of the first author’s lab notebook then revealed missing contemporaneous entries and raw data for key experiments.”
I don’t need to remind Retraction Watch readers what a huge, huge red flag this is. A high-impact paper gets published in Science. The results can’t be replicated. Then it turns out that not just any data, but rather the *key raw data and documentation*, are suddenly missing. Call me a cynic, but I can’t shake the suspicion of intentional data fabrication here, likely on the part of the first author. I very much doubt that Prof. Arnold was involved in data fabrication herself, but as she admits, her reputation, combined with her lax oversight of her personnel, likely contributed to the ease in getting the manuscript accepted based on flimsy (at best) or fabricated (at worst) results. I think this situation needs further investigation. It’s fine to commend her for freely admitting her mistake to the public, but a mere Twitter apology shouldn’t be the end of the affair.
We do not know a lot of background and we could just speculate. I agree a public apology is not enough. In the field everybody talks nicely about FA, I have not heard anything bad about her. The responsibility is not only of the authors but also of the reviewers and editors. A lot of times editors and reviewers are “intimidated” by the CV of the corresponding author and simply they do not review properly the manuscript.
I agree with Matthew. Being a professor myself, I know all too well how things are done in academia. Every supervisor has the full responsibility of what they publish and of the data coming out of their lab. I know big shots are busy, yet this should not be an excuse to allow (and reward) bad behaviour. It is too easy to just discuss the issue and blame the student who did the work. He or she is certainly to be blamed, but so is the head of the group, because she should always be in control of what comes out of her lab, period! Scientific integrity should always come first, fame and recognition second. Sorry Dr. Arnold, being a superstar does not give you a free pass for publishing everything and anything that has not been carefully double-checked and that is not of the highest standards. In fact, because of your status you even have a responsibility to lead the way for integrity and quality in science. There is already so much junk being published today, science (not the journal) really needs strong and trustworthy leaders!
I’ve worked in a number of different labs. All the faculty I have worked for had no time to look at raw data–in fact, I had a few look quite impatient when I tried to get their opinion on some. In the end while faculty should be responsible for the quality of these data coming out of the lab, they simply cannot or will not take the time to insure this, and its up to the data generator (grad student or post-doc) to make sure that its correct and in publishable form. In light of the fact that this is as important then any “management” that a faculty member may provide, or any grant money they provide, I think its unethical that data generators are asked to work so hard for so little money, while faculty retain their great salaries and job security. As long as this situation remains, where data generation is a relatively crappy job, this will continue to happen.
This happened when I was in grad school 20+ years ago and still happening now.
NMH, this is terrible! It seems you have worked with/for a few cowboys there. As a faculty, I have always looked very carefully at the data generated by my students and Post-Docs, asked them to redo control experiments, and always make sure their data is reproducible and their experimental parameters under tight control.
I cannot even imagine how a faculty member/head of a group could not eagerly want to spend the time to look at the raw data and ensure everything was done properly. This is just basic proper science, and that’s what students have to learn in order to become integer researchers. In the end, either you do science (using the scientific method the proper way, which require to always be in control), or you don’t. If you don’t this is pseudo- or junk science.
The first author is no longer working in the lab and the senior most author-main person responsible for publication bailed out just with a mere apology. This is a trend of politics, not science. छोटा आदमी पहले पिसता है।
What is a big deal here? Of course, she knows how to manage a lab. Sometimes, a bad apple can create a trouble to the PI.
It is fine and commendable to retract a wrong article and apologize, but this retraction NEEDS to come with a full disclosure of what the problem exactly was. Which data was irreducible and in which of the figures and panels? What was the nature of the problem(s)? Technical, interpretation or data fabrication or something else? Who was responsible for that?
Without a full transparent disclosure of the facts and of the responsible people involved, this retraction is insufficient. A retraction consisting of just five lines and just generic information is frankly a joke.
I have to disagree here.
The article was retracted in its entirety. The authors are not confident in the data.
By retracting everything after discovering something was amiss, the authors do the field a service by not letting other researchers waste time and resources.
In situations like these, a full accounting of what went wrong can take months, especially if there are disputes and an independent committee is involved.
Better to retract, determine what happened, and publish valid results (if any) at a later time.
If it is just the one paper, then I think Prof. Arnold’s response is better than the vast majority we have read about on this site. If other problematic papers appear (and you know people are looking), then it’s a different story. I would like to assume the best for now.
I’m confused by why people are blaming the PI so much in this situation. I am a PI and my lab group size is about 10 people typically (mix of graduate students and undergraduates). @NMH: as a PI, I agree, I am always willing to look at raw data when my students want to show it to me.
More broadly, though, what is a PI’s responsibility supposed to be? Should they look through everybody’s lab notebook, at every relevant page, before a paper is submitted? What if someone fakes their lab notebook… maybe the PI should install cameras in the lab and check the video records, to make sure the data really is being taken. Yes I’m being sarcastic. I just think that one needs to trust lab members to do the work, and be ready to engage in discussions about raw data from time to time, but in general err on the side of trust rather than suspicion.
And on the other side, yes I agree this looks like a case of fabrication. But the PI in this situation may not want to say that publicly until there has been an investigation. We can speculate in the comments section of this blog, but as a PI I think the best response is to retract quickly (given the results are so much in doubt) and then figure out the details of what went wrong, taking whatever time is needed for that. Yes, it would be nice if eventually the retraction notice has more information about what went wrong, once it comes to light. But I would rather see a quick retraction without details, rather than a lengthy delay to get all those details.
I strongly disagree with the value of a quick and minimalist retration. The retraction notice at this point should clearly say which data (figures and panels named specifically) was not reliable and why. Which raw data were missing? Were they lost or never collected? That must be knowledge existing at this very time point. I agree that maybe a further investigation might be needed to assign blame and that might not need to be public, but at least the factual reasons for the retraction should be fully disclosed.
My take? This was done on the 2nd of January (vacation time) without any details in the hope that few would notice and a brief retraction woul be accepted and forgotten. It seems to have worked fantastically. I do not expect any further information to come out because it clearly seems that the authors and the journal wanted to get this over quickly. I am pretty surprised on how everyone has been praising the authors when a fully transparent retraction has not been provided.
I want to disclose that I love the work by Dr. Arnold and her lab and I am not implying any misconduct. I am just saying that a retraction should come with full transparent and specific disclosure of the errors in the manuscript, otherwise is pretty meaningless. I am pretty surprise that the Retraction Watch community is not criticizing this retraction for a lack of transparency.
@Postdoc: what you say makes sense. One could note which figures are based on the work that can’t be found in the lab notebooks (perhaps all of them) and be transparent in that way, without needing to decide how the mistake came about (accidental or on purpose). As far as “lost or never collected”, that is probably hard to determine quickly. If the person who supposedly collected the data says “lost” do you take them at their word? Unless there’s a quick confession of bad behavior, it may take a while to determine exactly what happened.
This is a great argument as to why labs should not have 10 people–the advisor cannot properly insure that people are doing their job. If you are being paid more, you need to work longer hours to make sure the data quality is what it needs to be. You probably expect your gard students work 50 hour weeks, so you need to do much more than that–you are being paid to do this.
All that it takes is another student to do the key experiments to verify prior to publication. It does not need cameras or extensive documentation. When students know that the key observations will be checked by a colleague, they are relieved because it indicates that their data will stand the test of time. That is the way to do it.
George, thanks for posting a balanced response and I agree with most of what you said. I also think Prof. Arnold did the right thing to retract the paper quickly, and that’s already better than many others would have done in this situation. Nevertheless, I still believe there is a real problem in science today when PIs do not take the time to do their basic duty of making sure everyone in their group follows proper practice.
I see too many of my colleagues who just don’t care, because they are too busy writing a grant proposal, or traveling to a conference. Still, this should not an excuse for lowering the bar. In the end it is about doing what is right to the very best of everyone’s ability. Mistakes can happen, thats understandable, but not doing your job to the fullest with the excuse that you are too busy with other things is just not cutting it.
The solution is simple, if you cannot make sure your data is rock solid and reproducible, and that your students do follow proper practice than hold it and do not publish! Especially not in a journal like Science which requires the highest standards. I think what Prof. Arnold is accused of here, and she recognised her mistake, is to have not spent the time to carefully go through the data and talk to her students (especially the first author) about the experiments and the results. In my book, there is no excuse for that, no matter how busy you are.
The last point I wanted to talk about, and that you referred to, is that of trust. Should we trust our colleagues and co-workers? Yes, but not blindly. We must still scrutinise and question everything. Mistakes can happen, people can be irrational for one reason or another, time constraints and the pressure to publish can push some of us to overlook good scientific practice. In particular, students are still apprentices (in a way we all are, right?) and they are bound to do mistakes. Should we trust that they will always get it right?
I would love to know whether they have all the relevant logbook entries and the raw data for their Angewandte Chemie paper from last year, which has the same first author…
Well, if they had tried to reproduce the experiments there would not have been a problem, would there?
We can solve part of the reproducibility crisis for lab studies by simply demanding that papers are based on reproduced results, even if only verified in the same lab. This was a sine qua non once upon a time. Still looks like a gold standard to me.
Absolutely! There is only one way to do science, the rest is BS. It is either done to the highest standards, or it is not science.
When you submit a paper to Science, you make sure it is correct. You have another student repeat the key findings. This is easily done and reflects the approach to verification and reproducibility. That this was not done points to gross negligence at the very least.
Well, these days, the raw data associated with published work is supposed to be kept in a safe place in order to refute any claims of misconduct. I would say that if a PI simply required lab members hand over the raw data (or copies thereof) associated with a given paper sounds reasonable. In this case, it would have shown up missing. PS: 10-15 people isn’t a problem in my book, so long as the PI is aware of what’s going on. 50, on the other hand…taht’s asking for trouble.
I appreciate the courage and respect Frances for retracting a high impact publication. There can be serious repercussions for not retracting timely, but I still sense more of a moral responsibility that has been exhibited. I have worked with a very extended lab and lab that exists across continents, so number of lab members that a Professor can be higher than what Frances has. If researchers can actually act genuine and try and reproduce their work, “non-reproduction” based retractions can actually see a lot bigger number !!!
It is a muddled circumstance, these alleged top researchers get huge entireties of award cash, procure scores of researchers who contend with one another to satisfy the manager who has no opportunity to take a gander at the information or the composed original copies since the dominant part of their time is squandered on flying and conveying a similar talk at various foundations. Let us bolster progressively youthful researchers with better thoughts and inclination to accomplish more than this purported setup, researchers.