Hundreds of dead rats, sloppy file names: The anatomy of a retraction

via PubPeer

It all started – as more and more retractions do – with a post on PubPeer, this one in November 2021. The comment was about a paper titled “Efficient in vivo wound healing using noble metal nanoclusters” that had appeared in Nanoscale in March of that year: 

Figure 5: There is an overlap between two images taken from different experimental conditions. I’ve added a version below with the contrast enhanced. It’s difficult to match the brightness perfectly, but all of the same structures can be matched between these two sections. Would the authors comment?

Vincent Rotello of the University of Massachusetts at Amherst, one of the corresponding authors, responded right away. “Thank you for bringing this issue to our attention,” Rotello wrote on PubPeer. “We take data integrity seriously and are investigating the origin of the image duplication.”

Rotello also immediately contacted one of his co-authors, in whose lab the figures had been constructed. “Please see the concern below published on PubPeer,” he wrote on November 4 in correspondence released to us by UMass following a public records request. “I examined the image more closely and their conclusion appears to be justified. Please look into this issue and let me know your response.”

The response to that was swift, too. “Yes, professor. I am contacting the first author who have completed that experiment urgently,” wrote the co-author, whose name was redacted by UMass, within a day. “I am concerned too. I will let you know what is wrong as soon as I have his answer.”

Rotello and his co-corresponding author, Xifan Mei, of The First Affiliated Hospital of Jinzhou Medical University in China, had submitted the paper on Oct. 7, 2020. The reviews – also made available to us after a public records request – came back on November 24. “I have carefully reviewed your manuscript and the reviewers’ reports, and the reports indicate that major revisions are necessary,” wrote an associate editor whose name was also redacted by UMass.

Neither of the two reviewers saw any problems in Figure 5, nor in any of the images that would eventually be flagged in other PubPeer comments. Rotello confirmed to us “that no issues were raised by the reviewers regarding the figures in question.”

One reviewer requested additional experiments, and after a reminder to send a revision on December 10, Rotello wrote back on Christmas Day to say those experiments were in progress. The associate editor had not set a deadline in the original response letter, but said they were extending the deadline to January 31.

The authors submitted their revision that day, and on March 4, the associate editor sent an email with good news: After another round of reviews, the paper was accepted.

It was published on March 8.

Shortly after Rotello responded in early November 2021 on PubPeer and forwarded the critique to his co-author, other comments appeared on PubPeer. It wasn’t just Figure 5 that seemed to have issues, but also Figures 1a and S12. All of them were part of the original manuscript, not the additional experiments the reviewers requested.

Meanwhile, the first author responded to the email thread and wrote to the editors on November 8 and asked for a retraction. Kuo Li,whose name is redacted in the correspondence but whose identity was confirmed for us by Mei, wrote:

I am writing to retract the manuscript (Efficient in vivo wound healing using noble metal nanoclusters, Due to improper grouping and naming methods, different types of pictures were not classified into different folders, but they were arranged based on memory, which lead to confusion in processing data. Several related pictures are disordered. After repeated verifications many times, I found that these strategies could treat rats’ wounds. If you could repeat this experiment, you would find the drugs indeed work. However, I was very sloppy about details that caused an irreparable mistake for displaying the images. For this, I deeply apologize and must retract the manuscript. All details will be re-examined. I apologize again for my stupidity and lack of rigor, but it will not happen again anywhere. Thank you very much for all of your help. I will learn my lesson strictly in the future.

Li, who left the lab a year ago, hasn’t responded to our request for comment. But Mei explained that Li “saved the images of the same tissue into different files after completing the experiment. The names of these files were not significantly different causing images from the same tissue to be reused.”

It took the journal – or more accurately, the publisher – more than a month to respond. On December 14, the Royal Society of Chemistry’s Publishing Ethics Executive – whose name is also redacted – wrote, in part: 

Thank you for your message below which has been passed onto me. In order for us to consider your request, please could you send me a detailed description of all the errors in the manuscript? We will then consider the most appropriate action to correct the scientific literature.

As the publisher we have a duty to maintain the scientific record and will consider a retraction only if: 

– we have clear evidence that the findings are unreliable, either as a result of misconduct (for example, data fabrication) or honest error (such as a miscalculation or experimental error)

– the findings have previously been published elsewhere without proper cross-referencing, permission or justification (that is, cases of redundant publication) 

– the publication constitutes plagiarism 

– the publication reports unethical research.

Li responded within 24 hours:

It was found the images were put in separated folders though they were from the same samples. I would like to find the correct classification, but because a long time has passed, the test record is very rough, so it is difficult to clarify the right data. There are serious errors in four pictures including Figure 1a (control and GSH-AgNCs in 1 hours), Figure 5 (control and GSH-AgNCs, DHLA-AgNCs and DHLA-AuNCs in H&E 200× ) and Figure S12 (control and DHLA-AgNCs), in the article, making it very difficult to proofread the manuscript. These figures were accidentally came from the same tissue.

Li included the problematic images, and then concluded:

Nanoscale is a very high-level journal, and such errors should not appear in the journal. To avoid adverse effects on Nanoscale, I decided to take the initiative to withdraw the manuscript (DOI: 10.1039/d0nr07176e). At the same time, I will learn from the wrong lessons and deal with the data seriously. Thank you very much for your precious time.

The publishing ethics team wrote back the same day to acknowledge Li’s message. But it was nearly two months before they followed up again. On February 8, while thanking Li for his patience, they sent a draft retraction notice. Li responded the same day with some suggested changes, and after some back and forth the notice was finalized and all of the authors agreed to it.

The notice ran on February 16:

We, the named authors, hereby wholly retract this Nanoscale article due to errors in the labelling and filing of the data, which has led to ambiguity over the correct data for each figure.

In Fig. 1a, there is overlap between the 1h control panel and the 1h GSH-AgNCs panel.

In Fig. 5, there is overlap between the H&E 200× control panel and the H&E 200× GSH-AgNCs panel, and between the H&E 200× DHLA-AgNCs panel and the H&E 200× DHLA-AuNCs panel.

In Fig. S12, there is overlap between all of the panels for the control and the DHLA-AgNCs.

The original data is available, but the order and naming of figures are not reliable. Since the other figures still support the conclusion, the data related to Fig. 1a, 5, and Fig. S12 will be redone and verified multiple times.

Signed: Kuo Li, Dan Li, Cheng-Hsuan Li, Pengfei Zhuang, Chunmei Dai, Xiangka Hu, Dahao Wang, Yuanye Liu, Xifan Mei and Vincent M. Rotello

Date: 16/2/2022

Retraction endorsed by Heather Montgomery, Managing Editor, Nanoscale

Despite the delays – all of which were on the part of the journal and publisher – three and a half months from a problem being flagged to a retraction is quite swift. That is one punchline of this story.

But there is at least one other. The study was designed to test a way to heal wounds – a method that Mei says has held up in other work. And that meant sacrificing a lot of lab rats for data that ended up in a retracted paper.

“We remembered that Kuo Li had to kill hundreds of rats and studied gold distribution at different times,” Mei said. “Because so many rats were killed at the same time, we felt very bad when we heard about that. It was very memorable and sad for these rats.”

Like Retraction Watch? You can make a one-time tax-deductible contribution or a monthly tax-deductible donation 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

4 thoughts on “Hundreds of dead rats, sloppy file names: The anatomy of a retraction”

  1. “memorable and sad for the rats” (!) Right. Sloppily conducted and mismanaged experiments like this are not only bad science, but wasting animals like this is completely unethical.

  2. These are Bik category II errors, though. It is certainly possible that someone at some point took a single image, made multiple crops of it, put them into badly named files and later mistook them for different samples. But it’s not such a common form of error compared to simply pasting in an image from the wrong sample. Bik suggests being quite concerned about a paper with multiple category II errors. I’m glad this one was retracted and not corrected.

  3. Three and a half months between the request for retraction and the retraction. If only there was some mechanism that the journal or publisher could use to Express to readers that there is some Concern about the paper… perhaps a “pronouncement of disquietude”?

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.