This is my question for PLoS Biology today. You may recall some time ago I spoke out regarding what appeared to be data of questionable integrity in paper at the journal. Last month I provided a follow-up to coincide with the 6 month anniversary of the paper’s publication, highlighting a complete lack of action taken in response to my findings. Well today, I got the following email from the editor…
Dear Dr. Brookes,
I am writing to let you know that we have issued a formal correction for the manuscript entitled “Effects of Resveratrol and SIRT1 on PGC-1α Activity and Mitochondrial Biogenesis: A Reevaluation” by corresponding author John Holloszy, published in PLOS Biology. The authors inadvertently used the wrong blots in three of the main figures resulting in duplications. You can follow this link to see the correction: http://www.plosbiology.org/annotation/listThread.action?root%3D78137
Thank you for contacting us regarding these issues.
Sincerely
The correction itself reads as follows:
The authors inadvertently used the wrong blots in three of the main figures resulting in duplications. The blots in question are PGC-1a and LCAD in Figure 1A, Cyto C in Figure 4C, and COXIV in Figure 6B. The authors and their institution have confirmed that these errors do not affect the interpretation of the results or conclusions of the paper. The authors apologize for any confusion caused by these mistakes and thank the readers for alerting them to these issues.
Please view the correct Figure 1 here:
http://www.plosbiology.or…
Please view the correct Figure 4 here:
http://www.plosbiology.or…
Please view the correct Figure 6 here:
http://www.plosbiology.or…
Ignoring the fact that the correction was issued a week ago, and they only just saw fit to tell me about it, I guess that clears everything up then!
Except…
(1) The problem with Figure 4A was not duplication, but splicing. Some panels of a blot were spliced (in one case, every lane of the blot) but the supposed loading controls were not spliced. As such, these controls could not possibly have arisen from the same gel or membrane and are therefore not proper loading controls. This issue is completely un-resolved by the correction.
(2) The correction claims that the authors “inadvertently used the wrong blots in three of the main figures”. To me, “wrong blot” implies the whole blot. Not part of a blot. Not one lane of a blot. The whole thing. In the case of Figure 6 this may be an adequate explanation – the entire blot (all 4 lanes) was wrong, so the whole thing has been replaced in the correction. BUT, in the case of Figure 4C, the substitution was not a whole blot, it was a single lane of a blot. How exactly does one “inadvertently” use the wrong band in a blot?
Just to be clear, the phrase “inadvertently used the wrong blots” is completely inadequate as an explanation for what actually gave rise to the figure – duplication of a single band between blots via un-disclosed splicing.
In the case of the LCAD blot in Figure 1, again it was not simple duplication of an intact blot – the lanes were shifted over by one position. Is “used the wrong blots” an adequate explanation, for the practice of using the same blot panel and shifting the lanes over by one position (not to mention adjusting the saturation) to represent different experiments? Is “duplication” an adequate explanation for what happened here, given the manipulation that took place to give rise to the different images.
(3) The corrected images are interesting to say the least. Each image is presented as a TIF, which is not particularly problematic, until you look at the horizontal/vertical alignment. To do this, open a Figure (say Fig. 4), hold down the CTRL key and roll the mouse wheel to magnify the image. Then scroll vertically so one of the features which should be horizontal (e.g. a graph x-axis) is aligned with the upper or lower boundary of your computer window. Notice anything funny?
That’s right- these images are not aligned, as they would be if they’d been generated from a software package such as PowerPoint or Illustrator. I did this for the graph at the bottom of Fig. 4, to demonstrate…
…but it’s the same for every part of the Figure, including the bottom edges of all the blot images. How did these figures come to be rotated ever so slightly? My speculation is that they were scanned in from a paper copy at some point, and someone mis-aligned the paper in the scanner.
So here’s the problem… Why, when the integrity of these data was being questioned, would the authors choose to use such an arcane method of figure preparation which raises more questions than it answers?
Why did it take 7 months for the journal to do anything about this? Do they honestly expect their readers to accept such a totally inadequate explanation from these authors? Do they really expect us to buy that this “does not affect the interpretation of the results or conclusions of the paper”? Do they think I was born yesterday?
Naturally, a response email calling out the above issues has already been sent, and posts to PubPeer, PubMed Commons and the PLoS Biology page itself, are in progress.