Meet the woman who’s tracking down systematic research fraud

An interview with Jennifer A. Byrne

The topic of research fraud is a serious – and growing issue. In this article, we interview Professor Jennifer A. Byrne opens in new tab/window about her work in identifying systematic fraud, the software she’s helped develop and the pioneering work she’s been doing to promote a better appreciation and regard for the importance of a “clean” body of research literature.

"We need to value the process of cleaning the literature as much as we value the processes of adding to it."

1. Tell us a little about your background and research interests. I’m a molecular biologist and a cancer researcher. My research interests include studying the functions of specific genes in cancer, investigating the genetic basis of childhood cancer predisposition, and studying the operations of cancer biobanks.

2. How did you begin your work on (systematic) fraud? This started by accident, when I read five papers about a gene that my team had identified years before. These papers were very similar, even sharing particular nucleotide (or gene) sequence reagents. I could also see that the same reagent was being used in different ways, which couldn’t be right. Further analyses revealed that some reagents were wrongly identified, meaning that some reported results were impossible. When I realised that many other papers had these same types of errors, I fell into a strange new scientific reality, where I’ve been ever since.

3. You’ve been called an “Error sleuth” – is that how you see yourself? “Error sleuth” certainly reflects my focus on detecting errors in publications and manuscripts, but this is actually more of a means to an end. We think that particular errors might be more frequent in, or even characteristic of problematic papers, and it’s the affected papers that we need to find- the errors are the clues that could lead us there.

4. What has been the most surprising element of your discoveries thus far? Oh, where to begin… the sheer number of papers that are affected by these errors, and the huge number of incorrect reagents that we’re discovering, continue to take me by surprise. We started by analysing papers that rely on a particular form of nucleotide sequence control that shouldn’t target any gene. These control reagents are intended not to work, yet we identified control reagents with predicted gene targets. I imagined that this would be the main error type that we would encounter going forward. However, as we analyse more papers, we’re finding different errors to be more common, such as reagents that are supposed to target a specific gene, yet don’t appear to target any gene at all. These reagents are supposed to work, yet without a target, they can’t. It’s really hard to understand how papers can report results from reagents that can’t work. It’s the same as expecting your car to start without petrol in the tank.

5. Tell us about the Seek & Blastn software that you have developed… What does it do? Early on, when I didn’t know what to do next, I did what we often do in science- I wrote to someone whom I’d never met and asked for their help. That person was Dr Cyril Labbé in France. He had written a tool to detect automatically generated papers, and he helped me to apply that tool to the papers that I was studying. As a computer scientist, Cyril also saw the possibility to write a new tool that could identify, extract and fact check nucleotide sequences in publications, so he wrote the Seek & Blastn tool to do exactly this. The tool is freely available at http://scigendetection.imag.fr/TPD52/ opens in new tab/window for anyone to use.

6. I understand that there are plans for at least one journal to start piloting the software you’ve developed. Tell us more! (Do you think this is the first of many, for example?) We have a journal which has agreed to pilot this software, which is really exciting, and we hope that this will be the first of many. Screening procedures need to be widely adopted across different publishers and journals, to effectively deter problematic manuscripts.

7. You’ve likened the responsibility for looking after the “ocean” of research literature to being environmentally conscious. Would you say that this kind of activity has become more highly valued nowadays? I believe that there’s a growing awareness of the literature as our research environment, and the need to take care of it, as opposed to just pumping in more and more information. There are signs that we’re moving in the right direction, but as with our responses to climate change, we’re not moving fast enough.

8. What are your plans for the next step(s) with your work on fraud and Seek & Blastn? We currently have funding from the US Office of Research Integrity. As a consequence, we’re improving and applying Seek & Blastn to as many papers as possible, manually checking the results, reporting more papers with errors to more journals, and writing our next manuscript. We’re also applying for more funding. However, as much as we need funding, we need researchers to apply Seek & Blastn to papers in their own fields, and to describe what they find to the scientific community.

9. You’ve said that we need to “talk about systematic fraud” and have applauded some progress that has been made in certain geographies. What other recommendations do you have for the scholarly community? In a broad sense, we need to spend much more time and money on maintaining the health of the literature. We need to value the process of cleaning the literature as much as we value the processes of adding to it. They are two sides of the same coin. And at an individual level, peer reviewers and researchers need to pay closer attention to the details in manuscripts and publications. This takes time, but nothing like the time and money that can be wasted from following incorrect leads.

10. How can publishers and journals help? This is the good news- there are so many ways. Simple changes, such as ensuring that journal and publisher contact details on journal websites are easy to find, and current, make a big difference to individuals reporting papers of concern. Journals can also help by reporting changes in manuscript submission patterns, and features shared by unusual proportions of submissions. Publishers can help by providing dedicated opportunities and support to editors who wish to share these issues. They can also invest in screening tools and integrity experts who know how to use them, and additional human resources to allow journals to promptly respond to and investigate reported concerns. New resources will clearly cost money, but as we know from medicine, prevention is far better- and far, far cheaper- than cure.