Biophysicists in India have retracted two crystallography papers describing protein binding sites following “concerns,” according to one retraction note.
The last author on both papers, however, told us he believed the retractions were the result of “trivial errors.” Although one journal praised him in its retraction note for his “positive engagement,” he said the process left him feeling “disgusted.”
One paper, “Structural Studies on Molecular Interactions between Camel Peptidoglycan Recognition Protein, CPGRP-S, and Peptidoglycan Moieties N-Acetylglucosamine and N-Acetylmuramic Acid,” was withdrawn from the Journal of Biological Chemistry in August 2014.
The second, “Mode of binding of the antithyroid drug propylthiouracil to mammalian haem peroxidases,” was retracted from Acta Crystallographica Section F this month. Here’s the retraction notice:
The results in the article by Singh et al. [(2015), Acta Cryst. F71, 304-310] have been brought into question and the article is retracted.
Concerns have been raised about the results and modelling of the ligand in the distal heme-pockets of the structure reported in the article by Singh et al. (2015). Following further investigation, the authors and Editors have agreed to retract the publication. The original PDB entry 4qyq has been obsoleted. The Editors thank the communicating author for his positive engagement, and for addressing issues raised in a timely fashion, and apologise to readers for any inconvenience.
The JBC one-line retraction, as we’ve come to know and not love, offers no details on that retraction:
This article has been withdrawn by the authors.
But senior author on both papers, Tej P. Singh, a biophysicist at All India Institute of Medical Sciences in New Delhi, India, wanted to explain what happened. Here is his account of the JBC retraction, in which he states that the journal asked him to withdraw and then resubmit the paper with revisions — but, upon resubmittal, told him the work was of “low impact”:
In the JBC paper, there were two supplementary figures with several panels. In the figure on dot plots, one panel got repeated mistakenly. In order to prove that it was just an error, we submitted all the original data on dot plots with 19 panels which were accepted by the journal. In another supplementary figure on SPR sensograms, the legends of two panels got interchanged mistakenly.
Both these errors were corrected and we suggested to the journal to publish the erratum on these two aspects. However, the journal advised us to resubmit the whole paper with these corrections to go through the fresh review process. I accepted it. So, we resubmitted the paper with these corrections but we received very subjective comments that the work was of low impact.
It was rather strange because it was published in the same journal previously with good comments. It was a very irrational act but we decided to move on and submitted it to another journal.
Also it is hard to understand the rationality of this approach because the paper has already been published in a hard copy volume of the journal. Furthermore, these data with very minor errors represented a very small component of the paper. The main substance of the paper was structure.
We asked the JBC to respond to Singh’s account, and got the following response from editor-in-chief Martha Fedor:
I’m afraid that we are not able to provide the information that you requested because the JBC maintains strict confidentiality regarding the editorial review process.
The Acta Crystallographica Section F retraction, while also voluntary, was also due to a “minor reason,” Singh told us:
The second paper had even a more minor reason. Here a so-called reader who assumed something because elaborate discussion was not presented due to the nature of the journal on communications. When we pointed out, then the reply came that we did not make certain statements in the original version. I really felt disgusted and decided to withdraw the paper voluntarily.
Both these pieces of works are unrelated.
Editor-in-chief of the International Union of Crystallography (IUCr) journals, including Acta Crystallographica Section F, Samar Hasnain, says that the authors communicated to the journal that they would try to improve their sample quality and try again:
After peer review and publication, during which time all data were made publicly available, a reader voiced some concern about the interpretation of the data with respect to the assignment and placement of the ligand that was reported in this communication. The editor responsible for handling the manuscript then undertook a further review of the data and manuscript. Two further independent experts were also consulted to resolve the issue. The concerns raised and the specific technical questions identified were then posed to the communicating author. By return, the corresponding author provided additional details not documented in the paper and answers to questions. The author requested some more time to carry out further analysis. This was done in a timely fashion. After this it was simply agreed that the degree of uncertainty in the interpretation of the results was such that the paper should be retracted. A significant technical difficulty is recognized in that the researchers are dealing with relatively poor samples, perhaps due to the fact that the protein is a heterogeneous glycosylated material isolated from native biomass. They have communicated that they will attempt to improve the sample and diffraction data quality, and try again. We wish them every success in this endeavour.
IUCr has a policy of rigorous and fair review process consistent with being a professional scientific union where experts from the community ensure the scientific vigour and technical quality of papers that are published. We are grateful to our readers and authors who engage positively to resolve any problems that may arise due to the demanding nature of the samples, technical limitations or limited data.
Singh says he has resubmitted both papers to “better journals.” He concluded,
Doing science should be a matter of pleasure but these things are bad in taste. Both these cases were based on trivial errors.
Like Retraction Watch? Consider supporting our growth. You can also follow us on Twitter, like us on Facebook, add us to your RSS reader, and sign up on our homepage for an email every time there’s a new post. Click here to review our Comments Policy.
Off-topic: I like the animated gifs used as covers in some sections of Acta Cryst.
Crystallography jargon follows.
Looking at the structure from the Acta F paper, it has some bad statistics, and probably could benefit from additional rounds of model building, but there’s nothing obviously problematic. The ligand in question has borderline electron density in two of four protein molecules and poor/no density in the other two, but nothing jumps out as particularly dubious if the interpretations are derived from the ones that have better density.
I’d like to see difference maps for all four, following some simulated annealing or otherwise bias-reducing refinement. This would give a bit better sense of what the data actually supports, but based on what’s available in the PDB, I don’t think I’d throw that structure out just yet. That said, the editors are deciding based on more information than just what can be pulled from the PDB.
I should clarify – the model does have some serious geometry problems which make all interpretations of the data difficult. I don’t mean to brush off the importance of refining a structure well before starting to infer the biological and chemical properties of the protein.
Reading through verbose statements of various involved persons, I sometimes get the feeling that the one-liner retractions by JBC are getting closer to the truth, and are in fact all the information the readers need to know.
About the JBC paper, I know the above is just one side of the story.
But in general, I guess it isn’t clear when a mistake should be retracted or simply corrected. Surely if the problem was central to the paper’s acceptance, I can understand why it should be re-reviewed. But with non-fiction books (i.e., textbooks where many of the authors of papers originally learned from), errors happen all the time. They get corrected, re-published and sometimes even as a new edition…
There is no doubt the ligand is there in molecules A&B, but you cannot see any electron density which would allow you to unmbiguously assign which way up it goes. Personally – I think they’ve got it in upside down – I’d flip it 180º around the S1-C3 axis and point the aliphatic tail towards the hydrophobic patch (near Phe 381). But I’d have no data for this.
The ligand electron density for molecules C&D is all but non-existent.
I’ve just rotated it.
Flip and it fits better – no clashes with the heme, instead you get a hydrogen-bond.
I agree – there’s a lot of anisotropy too, especially in the sulfur. Possible it’s bouncing between a couple of substates, though there’s no way you could confidently model that with the given data. I think some conscientious rebuilding would improve the maps enough to at least know better what the data does and doesn’t support.
I agree with David Briggs on the ligand density interpretation. It makes better sense to have the carbonyl close to the HEME carboxylic acid group and the propyl will be largely disordered. Given the ambiguity of the density, it could be something else too. It seems reasonable to withdraw and publish when better data is availbale.
Given “The second paper had even a more minor reason”, I wonder how the problem from the 1st paper is less minor.