My take on the NCBS paper being retracted, and the polarised conversation that has erupted around the incident, is here. The following are some points I’d like to add.
a. Why didn’t the editorial and peer-review teams at Nature Chemical Biology catch the mistakes before the paper was published? As the work of famous research-fraud detective Dr Elisabeth Bik has shown, detecting image manipulation is sometimes easy and sometimes hard. But what is untenable are claims by some scientists, and journals as well, that peer-review is a non-negotiable requirement to ensure the scientific literature remains of ‘high quality’. Nature Chemical Biology also tries to launder its image by writing in its retraction notice that the paper was withdrawn because the authors could not reproduce its results. Being unable to reproduce results is a far less egregious offence than manipulating images. What the journal is defending here is its peer-review process.
b. Nature Chemical Biology continues to hold the retracted paper behind a paywall ($9 to rent, EUR 55.14 to subscribe to the journal for a year). I expect readers of this blog to know the background to why paywalls are bad, etc., but I would have thought a retracted paper would be released into the public domain. It’s important for everyone to know the ways in which a paper was flawed post-retraction, especially one that has commanded so much public attention (at least as retractions go). Unless of course this is Nature Chemical Biology acknowledging that paywalls are barriers more than anything else, and the journals’ editors can hide their and their peer-review’s failure this way.
c. The (now retracted) Arati Ramesh et al result was amazing, etc. but given some social media conversations are focused on why Ramesh didn’t double-check a result that was so significant as to warrant open celebration once the paper was published, some important background info: the result was great but not entirely unexpected. In April 2020, Jianson Xu and Joseph Cotruvo reported that a known riboswitch that bound to nickel and cobalt ions also had features that allowed it to bind to iron. (Ramesh et al’s paper also cites another study from 2015 with a similar claim.) Ramesh et al reported that they had found just such behaviour in a riboswitch (present in a different bacterial species). However, many of the images in their paper appeared to be wholly manipulated, undermining the results. It’s still possible (I think) that someone else could make a legitimate version of the same discovery.