Fear of being scooped is fuelling the replication crisis in research

It was the nightmare moment that every junior scientist dreads: having found the result you have been searching for throughout your PhD, you learn that someone from another laboratory has the same finding and is about to publish.

Because the time from discovery to publication can be several months, or even years, researchers in the biological sciences always risk being scooped. However strong the research, the existence of similar studies will lessen a paper’s novelty and perceived impact in the eyes of journal editors, increasing the likelihood of rejection. And for an early career researcher such as myself, publications are pivotal for job applications and fellowships.

The proliferation of preprints during the Covid-19 pandemic – more than 39,000 were published on bioRxiv in 2020 – has provided an opportunity to reset this dynamic.

Traditionally, conferences are where unpublished work is discussed, and conference presentations can predate publication by several months. But conference audiences are necessarily restricted – and the fear of being scooped can be a disincentive to present anything that isn’t already in press. Preprint servers provide a new means for sharing and discovering research before it is published – and, potentially, for coordinating submission.

My work with Natalee Newton from the University of Queensland concerned a new method for determining 3D reconstructions of mosquito-transmitted flaviviruses, which include dengue, Zika and West Nile viruses. I had been solving structures of viruses using cryo-electron microscopy under my research group leader Fasséli Coulibaly at Monash University, and I teamed up with Natalee last year to prove that the chimeric viruses she worked on with her laboratory leader, Daniel Watterson, looked identical to the pathogenic viruses. This could potentially accelerate the development of vaccines for novel flaviviruses, which can emerge from animal reservoirs and cause epidemics.

At the same time, on the other side of the world, Max Renner, working with Gavin Screaton and Jonathan Grimes from the University of Oxford, was also studying the 3D structure of a close relative of the Zika virus that had not been characterised before. In his case, the focus was on viral maturation, the process by which newly produced viruses from infected cells become capable of infecting further cells.

Using different approaches and working without knowledge of each other, we had converged on the same unexpected result. But when I read Max’s preprint on bioRxiv, my coauthors and I were devastated.

A few emails and a video call later, however, we came to agreement. We would submit our research together, even though it required Max and Jon to hold off on submission until we had completed our manuscript.

Both teams believed this to be in the best interests of science. Biological research is not immune to chance and sometimes an experiment yields a result that cannot be replicated. But if two laboratories converge on the same answer, the probability of the result being random drops significantly. While most laboratories replicate their results with multiple repeats, cost considerations mean it is not always possible. In the case of cryo-electron microscopy, experiments cost thousands of dollars and time on the microscope is precious. So rather than being redundant, our experiments had validated each other’s finding in different viruses and strengthened the result of both experiments.

But coordinating publication is not always straightforward. Many journals do not have clear mechanisms for co-submission and do not sufficiently support the model. Our papers were treated as separate, independent studies and went to different editors and reviewers – which led to very different assessments of the quality and potential impact of each.

After experiencing several rejections, my group decided to split our paper in two, to streamline the story of each. We submitted one of them, alongside Max’s, to Science Advances. But while ours was accepted by one set of editors and reviewers, Max’s was rejected by another. Despite several letters from both research groups to the editor challenging the decision, we were unable to change the outcome.

Eventually, our remaining paper was published with Max’s in Nature Communications – and they have been accessed by thousands of readers since publication. Nonetheless, if this collaborative model is to be nurtured, more scientific journals need to support and have guidelines for reviewing and accepting joint submissions.

Because of commercial patents and performance-based funding, research is a competitive process. But if we are serious about developing robust theories and evidence-based medicines, collaboration also needs to be better enabled – especially now that preprint servers allow us to see more easily what others are working on prior to journal publication, and potentially approach them about co-submission. A greater willingness on journals’ part to recognise preprints as legitimate claims of discovery and to support the submission of work that has been shared as a preprint would also be a step forward.

For Max, Natalee and I, publishing together improved both our papers and added credibility to the results, identifying new drug targets for flaviviruses that could save lives. And, judging from the warm reaction to my tweet thanking Max, a shift towards collaborative approaches to publication would be welcomed by the scientific community.

Josh Hardy is a research officer at the Walter and Eliza Hall Institute of Medical Research in Melbourne, Australia.