Some readers may recall this post from 2015, which details problems with a natural product isolation paper in PLOS ONE. The compound, named Xinghaiamine A by the authors, was. . .well, let’s say extremely unlikely, and I think anyone who looks at the structure in that earlier post will agree. And that spans several levels of unlikeliness – at some point, there’s no more forehead for your eyebrows to crawl up, you know?
And that was before people starting having a look at the NMR spectra that were presented as evidence. You can see a blowup of one 2D spectrum in that earlier post, and a middle-school student could tell you that it is a crude cut-and-paste job. Faked, doctored, altered: it’s junk, and after seeing it no one competent could believe the paper’s structural assignment or vouch for the credibility of its authors.
A clear-cut case! I and others alerted the editorial staff at PLOS ONE. My own email laid out the situation, and concluded like this:
. . .I cannot see any other good explanation for the state that these spectra are in – they appear to me to not only have been faked, but faked quite incompetently.
Please have a look at this situation and see what conclusions you come to. And please, if possible, check to see who reviewed this manuscript for you. I cannot imagine a synthetic organic or natural products chemist letting the proposed structure in this work pass, even were the spectra not manipulated. But the rather egregious doctoring of the data makes this even harder to defend from a reviewer’s standpoint. Thanks very much!
Well, to my surprise, I had a reply from the journal just recently, informing me that the paper has indeed been retracted and apologizing for the delay. Retraction Watch noted this here. According to the retraction notice, the journal requested the raw data from the authors, who did not provide it and instead requested that the paper be withdrawn. Why this took three years is what both the Retraction Watch people and I are wondering.
As for those NMR spectra, the notice says that the corresponding author “indicated that spectral images from NMR analyses of different purification batches had been combined in error when deducing the reported structures“. Why yes. It is indeed an error to use image-editing software to cut and paste peaks and assemble a spectrum. If your NMR spectra from different batches do not agree with each other, you don’t get to glue the pieces together into what you think one should look like.
The corresponding author is at the Dalian University of Technology in China, with other authors coming from Dalian Medical University, the Dalian Institute of Chemical Physics, and Myongji University in Korea. Let’s hope that standards are higher there than they were in 2015.