Science retracts a paper by Dana-Farber CEO Laurie Glimcher. Dana-Farber claim to have started an investigation in 2021. The PubPeer thread on this paper has been active since 2014. The timeline doesn't seem very flattering...
Nature Medicine recently corrected a figure after my PubPeer comment and email. The "corrected figure" contains a different version of the same error. Current APC $12290.00.
Five different papers using the same image of nanoparticles. Are they selenium, aluminium, or silicon based nanoparticles?
Who supplied them?
No response on PubPeer so far...
Usual suspects: MDPI, Frontiers, Scientific Reports
I have posted a new preprint: 16% of papers with research images in the journal Toxicology Reports contain duplications. Thanks to everyone who helped!
Finally retracted by MDPI! I posted this in 2021. The paper includes lots of duplicated rat wounds, both within the paper, and shared with other papers. The authors initially disputed these findings.
@MattNachtrab
It wasn't Elisabeth who did any of the "shaming" (if there was any). She just pointed to the errors. But for the record, all the authors on a paper should take responsibility for the data in it. You're just calling for people to stop pointing out errors 🤡
In just one paper, annotated by
@ImageTwinAI
😐 If Springer Nature or Oncogene wanted to fix this, it wouldn't be difficult, so presumably they don't care. Cited nearly 200 times
Following my commentary on PubPeer and blogging Francis Hornicek at
@univmiami
has three papers retracted, so I've made a new episode of Science Police! (With the full intro back by popular demand)
Scientists from Harvard gain ethical approval to aspirate bone marrow from healthy volunteers and cancer patients, then mix up the images during publication. Surely this is an enormous ethical transgression. I would be furious if I volunteered!
MDPI in 2022 and T&F in 2023, authors pretend to do SEM but stole the image from a chemical suppliers' website. Perhaps they had to steal the image because apparently their SEM only has magnification of x10??
If you have a few minutes today, please read my latest blog 🙏 Researchers at MSK, Columbia, Penn Medicine, and more, caught in the kind of fraud that we would laugh at if it was published by MDPI.
I have just noticed a correction for one of the DFCI papers. New images have been sourced to replace these overlapping areas, the full text of the correction follows.
Does Science have a policy of retracting obviously Photoshopped figures? They never responded to my enquiry in January about these neurons (published by a different team).
Author response:
"...note that 10% editing is acceptable as long as we didn't modify the significant features"
I wasn't aware of this rule, so everyone please stay informed!!
I don't want to only post about DFCI, as I have many other projects going on, but this post really shocked me. I had no idea of the salaries of the individuals involved😲
A quote from a recent paper in BMC cancer about non-coding RNA: "Further information regarding the biological activity of SNHG1 in cancer is difficult to ascertain, due to the literature being corrupted by fabrication of papers for profit"
Should add that while these fake papers are [hopefully] ignored by most scientists, they do in fact get citations -- because they're citing each other -- and some do get aggregated into systematic reviews, particularly in the medical literature.
I asked GPT4o about this figure to see if it would spot the duplication between 3 and 4 (it does not, I added the red rectangles), although it did point out a contradiction in the figure legend that I did not see. Are the images samples of lung or stomach?
Are we expecting another three years for Laurie to deal with more recently identified papers? How about this mess which Glimcher co-authored in Neuron, which I pointed out in January? To be retracted in 2027?
Another concerning Norman E. Sharpless error. How to show a remarkable reduction in bioluminescence... Just show the Day 1 image again and label it as Day 10😎
@twillis209
Lots of different reasons, I think. Error in documentation, lost data, just thinks it looks nicer and doesn't really matter, outright fraud. This one involves Massimo Loda, so probably fraud.
It's good to see some progress, but I don't see how this is a sensible way forwards. I've sent hundreds of emails about papers to different journals, most of which have not been acted on. I'll be lucky to get an automated response.
STAT posted a story about nomenclature errors in the scientific literature, leading to a woman receiving an untested drug. I was peripherally involved because Tara asked me to help correct these errors in the literature.
Today's diagram, one already retracted (thanks to
@SmutClyde
)
@ImageTwinAI
makes these easy to discover (although I usually add the rectangles myself), the only limitation is the number of colours, and amount of time in the day
Several weeks ago, I annotated a series of papers on gastric ulcers with a large number of recycled and overlapping images. Today
@PLOS
retracted nine papers. I thought the methods in these papers were quite cruel to the rats, which were no doubt involved at some point.
A couple of weeks ago, I posted this diagram.
@DovePress
has now retracted the four they are responsible for 🥳 Any progress on your three contributions
@SpandidosP
?
Reported to
@HeliyonJournal
over two months ago, multiple overlapping images, but also...
Male ovaries and female testes 🤐
Authors tried to correct the male group, but still wrote about ovaries in the figure legend.
Also, I have seen authors express "regret" for missing an author's middle initial in a published Corrigendum previously (see link). No one saw fit to apologize for muddling patient's donated tissue here!
@MattNachtrab
@Proofig
@MicrobiomDigest
Matt, you are completely clueless on how to improve quality in scientific research. It is trivial to organize thousands of images in folders. A typical paper contains less than a fifty, which should be directly compared to the raw data before publication.
Fun game anyone can play: Choose a famous protein and a famous institution. Perform a scholar search. You will shortly find egregious cases of research misconduct. Here's one including Norman E Sharpless, until 2022 director of the National Cancer Institute.
"Bee venom: A potential natural alternative to conventional preservatives for prolonging the shelf-life of soft cheese ‘Talaga’"
Published by Elsevier's gutter Heliyon in April 2024, the images do not credibly depict the passage of time.
@MicrobiomDigest
Something interesting about this, Tirosh and Ploegh were contacted by the Washington Post about this figure and were quoted in that piece effectively saying there's nothing wrong with the paper... It was later updated.
@IvanOxygenLab
Journals should be forced to either integrate PubPeer or allow prominent (moderated) comments on all papers. Letters to the editor are too slow and frequently rejected.
@ImageTwinAI
Deserve an enormous amount of credit for developing a fantastic tool, and truly working towards their goal of improving the integrity of science.
Our mission is to increase quality and trust in science and research integrity and we would like to thank David Sholto (
@addictedtoigno1
) for working on that mission alondside us.
Further problems including extremely unlikely demographics and dubious preclinical justification see here:
@PLOS
How can this kind of silliness be uncritically published?
The correction does not acknowledge how the error was found. It does not explore why it happened, and it does not comment on how it could be avoided in the future. I think all would be relevant.
The first and second authors are affiliated with the company that manufactures ZingiVir-H. And yet... "Funding: The authors received no specific funding for this work", also "Competing interests: The authors have declared that no competing interests exist."
@mumumouse2
I occasionally cry when looking at images of mice in papers. Something also worth considering is that with the generation of antibodies and serum, the average biology paper can rip up the whole barnyard and it's often for nothing.
@MattNachtrab
@Proofig
@MicrobiomDigest
Remember that image duplication is only a very narrow category of mislabelling (i.e. where the same image appears twice). Scientists need to develop procedures to avoid all types of mislabelling, including those which aren't duplicates. Proofig is completely inadequate.
Published in
@Nature
: Four sets of duplicated images in three different experiments. Eighteen months since the authors accepted that errors were made... Curiously no correction.
Seems like GPT4o may be useful for analysing some types of errors in papers, with the cautionary note that it also misses some obvious problems, and gives superfluous commentary.
"In the version of the article initially published, three pairs of micrographs (in Fig. 5i and Supp. Figs. 3a and c) inadvertently contained overlapping areas. These figures have been amended with new micrographs and are available as Supplementary Information with this notice."
Nanoparticles go crazy again. At least this time the authors thoughtfully published an unedited version of the same image (described as something else) so you can be certain which bits were photoshopped
@MicrobiomDigest
@SciReports
I made a small contribution in that thread, this paper also includes long sections of plagiarised text... Seems like a lot of opportunities to catch this one 🤷