Exposing suspected research fraud with new software
Professor Jennifer Byrne, one of Nature Journal's ‘top 10 people who mattered in 2017’, delivered the 10th Annual Diana Temple AM Memorial Lecture at the University of Sydney on 9th October 2019. Professor Byrne shared her experience in studying the scientific literature, and her efforts in identifying and exposing flawed research papers.
In 2015, Professor Byrne first began her investigations through reading some cancer research papers, when she noticed several things that didn’t add up. She recognised a number of very similar papers studying a human gene that she had first reported. Professor Byrne then found that most of these papers described errors in the gene sequences that were used to produce the research results. In light of these errors, she was concerned that these papers actively encouraged further research.
“The risk is that researchers may base new work on these problematic papers, which could compromise research in the future,” she says.
Realising that similar types of gene sequence errors occurred in papers examining other human genes, Professor Byrne launched an investigation to understand the possible extent of the problem.
Through a collaboration with computer scientist Cyril Labbé at the University of Grenoble Alpes, France, the two developed new software called Seek & Blastn to detect errors in publications and manuscripts. Seek & Blastn extracts gene sequences from text and then fact-checks their claimed identities against an independent database. The Seek & Blastn program is freely available online.
The team have secured funding from the Office of Research Integrity to further develop and apply Seek & Blastn. They have now detected and reported hundreds of suspect papers, leading to 17 retractions and many more pending investigations.
How can we reduce low value and problematic research?
Pressure to publish, poorly regulated research environments, and inadequate peer review standards are all contributing factors to poor quality research, and rising numbers of retracted publications.
At the same time, researchers are developing methods to detect incorrect and problematic research. Most papers are now available online, which also makes it easier for readers to scan papers for errors, as well as image and data duplications.
“The digitisation of research publications means that it is easier for data to be fabricated, but it's also easier to screen publications for errors,”
Professor Byrne has also called for scientific journals to respond more quickly and effectively to errors reported by their readers. She emphasised that journals must be resourced to investigate many research papers simultaneously, in response to the use of screening tools such as Seek & Blastn.
“Journals need to more quickly and effectively identify papers of concern, to stop scientists from wasting time and money following up flawed results. We don’t have that time to waste,” she says.
