ResearchGate, the 4-million-strong academic social network, has just scored a major victory in its quest to turn the research process upside-down – its platform proved instrumental in the debunking of a major stem cell study, which on Friday resulted in a very apologetic roll-back from the Japanese institute that put out the original study.
The original study came out in Nature, the peer-reviewed scientific journal, earlier this year. It purported to show that it was possible to turn normal blood cells into master stem cells by dipping them in a mild acid solution. Stem cells hold great promise for the future of medicine, but there’s a lot of controversy around current harvesting techniques, particularly when embryos are involved, and this would have provided a terrific workaround.
Unfortunately, while the Riken institute’s research promised great things, other researchers found themselves unable to reproduce the results. One, a Hong Kong professor of regenerative medicine named Kenneth Lee, used his ResearchGate profile to publish his and his students’ findings, demonstrating a failure to get this technique to work.
There had been a good deal of grumbling about the Nature paper — notably from biologist Paul Knoepfler and on PubPeer — but Lee’s ResearchGate publication provided the final straw and on Thursday Riken admitted the original Nature papers contained errors. One of the key researchers there has now had her research suspended, and investigations are underway at both Riken and Nature.
“This is what I was dreaming of,” ResearchGate CEO Ijad Madisch told me. Indeed, Madisch has long advocated the advantages of “post-review” over “pre-peer review”.
Rather than the traditional system, where a paper is submitted to a journal like Nature and pored over for months by a couple of reviewers, he wants to see a system where all research is published openly and immediately – not only does this bring more eyes to the research, but it also means that unsuccessful results get as good an airing as the successful ones. That’s something that should save other researchers an awful lot of time.
Bill Gates likes this approach, which is why he has poured millions into ResearchGate. And, although the discussion around it occurred in more places than just ResearchGate, the stem cell episode provides a good deal of validation.
According to Madisch, researchers who have found it difficult to reproduce the results of published papers have generally had trouble getting as much attention as those behind the published paper – the press, certainly, are keener to trumpet a study that has been peer-reviewed than something that may come from a hater and that journalists feel unable to evaluate themselves.
The excitement around Lee’s activities on ResearchGate led the startup to speed up the completion of a new feature called Open Review, which is designed to make it easier for users to give open feedback to papers published on the network. The feature combines a structured feedback mechanism with commenting facilities.
“It’s open, real-time and transparent, and the authors can contribute,” Madisch said. “It forces the authors to provide more data. It’s important to understand that it’s not only to debunk research – we also want to highlight positive research.”
As the feature wasn’t available then, Lee began his reproducibility quest on ResearchGate’s Q&A system, but the sped-up release of Open Review made it possible to publish the efforts of Lee’s team in that format.
ResearchGate has found a clever contractual way to let researchers upload their papers outside the walled gardens of the traditional journals, and it’s growing fast. In its first 50 months, 2 million publications were added to the platform – the same number is uploaded each month now, and 700 datasets are going up each day. Here, for example, is the original Nature stem cell paper on ResearchGate, complete with a red note on the side to say it is “non-reproducible”.
“The next couple of months on ResearchGate will be so exciting,” Madisch enthused. “We have the reach, we have the engagement, we have this great feature, and the scientific community is accepting us.”
UPDATE (March 26th): Lee has published a Nature letter rejecting the study that ended up going on ResearchGate, claiming it doesn’t challenge the key points of the original study. Which is still not reproducible as of the time of writing.
This article was updated at 10.15am PT to more explicitly recognize the contribution of Paul Knoepfler to the debunking. His work was linked to in the original version, but not prominently. A mention of PubPeer has also been included.