How a typo in a catalog number led to the correction of a scientific paper — and what we can learn from that

Anita Bandrowski

Papers are corrected for lots of different reasons. In this guest post, Anita Bandrowski, who leads an initiative designed to help researchers identify their reagents correctly, describes one unusual reason for a correction — and explains what researchers can learn from the episode.

Last December, Tianyi Wang and her colleagues published a very interesting paper in Cell Metabolism on the potential link between a gene called JAK/STAT3 and breast cancer. It turns out that breast cancers become resistant to chemotherapy via a pathway that involves JAK/STAT3, and blocking this pathway can re-sensitize the tumor to chemotherapy.

But six months later, they corrected the paper — because of typos in the catalog numbers they’d used.

Here’s what happened: Wang and her colleagues had ordered an antibody to use in the study. Somewhere along the chain of graduate students, lab managers, and company representatives a single digit in the catalog number was switched from AF498 to AF497. Both antibodies actually target the same protein, but they do so in different ways and for scientific reproducibility it is important to understand how the protein is being targeted. Indeed, Hua Yu, the senior author of the study conveyed to us that it was the process of attempting to replicate part of this study by another lab member that led to the discovery that the wrong antibody was noted. The error was difficult to catch, but catch it the authors did, and kudos to them for correcting the paper.  

The authors also chose to correct the source of a mouse model they used. While the name and description of the animal was correct, the original paper said it had come from Jackson Laboratory — when it in fact was generated by a colleague.

It is rare to see a retraction for a reagent-related reason, although it has happened, and the Retraction Watch database includes categories for contaminated reagents and errors in materials. It is only slightly less rare to see a correction for that reason. So when we found the Cell Metabolism correction last month, we thought that we should celebrate these authors for their diligence — especially for the extra steps the authors took to help others. (More on that in a moment.)

So what can we learn from this correction? A lot, it turns out.

‘A STAR is born’

It’s very possible that one of the reasons these mistakes were caught was because Cell Press, the publisher of Cell Metabolism, requires that all papers include the STAR Methods format. This format is an incredibly simple three column table, but makes it clear which resources were used in a paper with one glance. The STAR Methods is sort of like a list of ingredients in your favorite recipe: It just shows you the list of the “stuff” you need to bake a pie, or, in the case of this study, figure out how to make cancer cells less resistant to chemotherapy. In this simple list is a very powerful means to communicate reagent information clearly. Many people worked to bring STAR Methods into existence, but one of the key members of this team was Katja Brose, the former editor in chief of the journal Neuron, who worked with a dedicated team of publishers at Cell to bring a shunned and nearly forgotten section of the paper, the methods, into the spotlight that they so rightly deserve.

There’s another reason to note this story. And that’s the Research Resource Identifier, or RRID.

RRIDs — part of an initiative that I direct — are a set of persistent unique identifiers, which point to “Key Biological Resources” as defined by the U.S. National Institutes of Health. Specifically, RRIDs are a set of numbers that mark antibodies, cell lines, transgenic organisms and software projects. These numbers are meant to persist, unlike catalog numbers, and always will point to the same antibody or mouse, as far as that is possible to assess in biology. Essentially, they mark the reagents that are more or less available to order, just in case someone wants to replicate some part of a previous experiment.  

RRIDs are now specifically requested by about a hundred journals, including Cell Metabolism. Cell Press is on the forefront of this particular sea change, adopting the RRID as part of their complete methods overhaul effort in 2016. The efforts of Cell Press, eLife, and many other highly respected journals, are indeed critical because, right or wrong, the community pays a disproportionate amount of attention to what goes on at these journals.

As a methods reviewer for several journals and a section editor, I run into various kinds of mistakes that make it into manuscripts. Recently, in reviewing methods for a neuroscience journal, I suggested to authors that antibodies used in the study should be tagged with RRIDs, and noted a few places where insufficient information was available to identify other reagents used in the study. The authors, responded by not only adding requested RRIDs, but they overhauled the methods section with new text and a lot more information that was not specifically requested. It is heartening to see that when authors pay attention to the methods, they themselves find mistakes and make great efforts to set the record straight.

Tianyi Wang looked up the RRID number that corresponds to the AF497 antibody in the RRID database before the paper was originally published, as required by Cell Press. In many cases, that act of looking something up in a database is enough to reveal a problem. If you used an antibody against one of cell coat markers under a catalog number 07-0403, for example, but someone in the lab at some point transposed a number incorrectly, writing 07-0402 by mistake, you would either get no results from the RRID system or you might get the information for a very different antibody back — and you would be alerted that something may be wrong. In some cases, the author believes his or her notes and registers a new antibody under an erroneous catalog number. Then, it becomes the job of the RRID project’s antibody curators to track down this reagent, sometimes calling the company, searching through older literature, or asking the author to take a picture of the bottle.

However, in Wang’s case, the catalog number matched a very similar product, so it was not so easily caught, but an attempt by someone else in the lab to replicate this study revealed the error. The antibody RRID has now been corrected, thanks to Wang, along with the catalog number.

This particular manuscript is not the only case in which we have seen this kind of author engagement. “It is not that authors try to mislead readers, they simply seem to pay little attention to the methods section of the paper once it is written,” says Maryann Martone, the editor in chief of Brain and Behavior, who has had several cases in which she or the authors caught rather serious omissions or significant errors while simply searching for RRIDs.

The RRID is something that seems to help draw attention to methods, sometimes revealing other problems, hopefully before the paper is published. Wang and colleagues are unlikely to be the last researchers faced with the “which antibody was that again?” question and they will likely not be the last to have a moment of “uh oh, I think that we wrote the wrong number in our paper.” However, the hope of those of us who deeply care about methods is that many of these sorts of “uh oh” moments happen before the paper is published — instead of after.

Like Retraction Watch? You can make a tax-deductible contribution to support our growth, follow us on Twitter, like us on Facebook, add us to your RSS reader, sign up for an email every time there’s a new post (look for the “follow” button at the lower right part of your screen), 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 team@retractionwatch.com.

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.