Retraction Watch

Tracking retractions as a window into the scientific process

Here are the 10 — yes, 10 — reasons PLOS ONE retracted this paper

without comments

Screen Shot 2016-01-19 at 10.50.25 AMPLOS One is retracting a paper for overlapping with a Wikipedia page. And for containing material lifted from other sources. And for “language errors.” And for insufficient evidence that authors found the pathogens floating around in hospital air that they claimed to find.

The instances of plagiarism are a “huge problem,” each “enough for retraction on its own,” Jonathan Eisen, a microbiologist at the University of California, Davis, told us. Eisen, who posted several comments to the paper after its publication in October, added that the paper was “simply not technically sound.”

The paper — which even contains a spelling error in its title, “Metagenomic Human Repiratory Air in a Hospital Environment” — describes a gene sequencing method to screen the air in hospitals for pathogens. The retraction note lists 10 concerns with the paper:

After the publication of the article, readers raised concerns about several aspects of the presentation as well as the methodological aspects of the study and the conclusions drawn. Follow-up with the authors and consultation with editorial board members, identified the following concerns:

  • The manuscript was copyedited on submission at the request of the editorial office yet language errors were reintroduced during the evaluation process. The quality of the language of the published article is not in line with the criteria for publication in the journal.
  • Fragments of text overlap with text from the “Hospital-acquired infection” entry on Wikipedia:
  • Table 1 was reproduced without attribution from: Julia S. Garner: Guidelines for Isolation Precautions in Hospitals Hospital Infection Control Advisory Committee. 01/01/1996,
  • Table 2 was adapted without attribution from: Pasanen, A.L., A review: Fungal exposure assessment in indoor environments. Indoor Air, 2001. 11(2): p. 87–98,
  • 87.7% of sequences in the MG-RAST dataset are noted as failing QC:
  • Evidence that pathogens were detected in the samples is not sufficient.
  • Data from MG-RAST have not been validated, either via other bioinformatics methods or independent approaches such as PCR or culture. How the potential non-specificity in the matching to taxons was addressed is not clear.
  • For certain genera and species, metagenomics abundance is compared to CFU/mL. No information is provided on how the different species on the agar culture plate were counted or confirmed, or how the normalization steps in library preparation and sequencing were accounted for when measuring metagenomics abundance.
  • Whether the results for PhiX174 originate from library preparation or contamination from elsewhere is not reported. The human endogenous retroviruses that are reported are likely reads from the human genome and should be classified as such.
  • The accuracy of the taxonomic assignments for the metagenomics reads In Table 7 is unclear. Several of the bacteria listed should not be classified as pathogens.

The Editors consider that the concerns outlined above compromise the integrity of the work and the validity of the conclusions reported and as a result retract this publication.

The paper was published in October, 2015, and has not been cited, according to Thomson Scientific’s Web of Knowledge.

Eisen had concerns about the paper immediately, he told us:

When I saw that paper I was truly dismayed at how many problems it had.  My 1st concern, which is in my 1st comment on the PLOS One site, was simply that it did not cite the other papers on the topic, which is a very bad sign.

Here’s Eisen’s first comment, from October 9th — just over a week after the paper was published:

There are a large number of previous papers where people used DNA sequencing to look at microbial communities in a health care setting that were not cited here. This includes this paper from Bokulich et al.… and this one from Brooks et al. http://www.microbiomejour… and this from Kembel et al.… and a variety of others.

The first author responded, in part:

Actually we really didn’t see the first paper, if we know it we’ll definitely refer to it. However, we would only refer it for Shewanella. This species is somewhat difficult to culture. We detected out it in hospital, still not assure where it came from and whether our results were reliable. Now a previous paper had detected it by 16S method. That would greatly improve the credibility of our work. So if we would definitely refer to it if we were seeing it. However, we possibly would not refer it for DNA method or ICU cleaning since we should notice the details of research.

And another comment from Eisen, also posted October 9th:

The authors do not present any evidence that they have detected actual pathogens in these samples. At best they may have been able to say that they have sequences that are related to sequences from pathogens. But in most if not all cases, such sequences would also be realted to those from non pathogenic strains of the same species.

The first author responded, in part:

Definitely substantial amount of pathogen is conditionally. And what important in our paper is that we concentrated on the network and the speed of the screen method, then combined with other methods for further study if necessary.

Two weeks after the paper was published, PLOS posted a note — basically an expression of concern — on the paper:

Several readers have raised concerns regarding the reporting and language quality in this article. Language quality was noted on submission and the manuscript was copyedited before peer review, but errors were later reintroduced. PLOS staff are looking into how these errors occurred and following up on the concerns raised about this article.

Eisen is a member of the PLOS Biology Editorial board, but said he’s not aware if that had a direct influence on the retraction of this PLOS ONE paper. We asked him if he felt these problems could have been caught in peer review:

Unquestionably some of them should have been caught.  I firmly believe in “publish everything that is technically sound” which is the PLOS One approach. But this paper was simply not technically sound and the technical problems really should have been caught in review. But not all technical problems get caught in review and that is OK.  This paper also has problems in other areas such as lack of attribution for material in the tables.

We’ve reached out to last authors Fei Xiao, who works at Beijing Hospital, and Geng Tian, who works at Tsinghua University. We were unable to find contact information for first author Yi Yu Lai, also at Tsinghua University.

Hat tip: WvSchaik and Kerry Grens

Like Retraction Watch? Consider making a tax-deductible contribution to support 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.

  • Post a comment

    Threaded commenting powered by interconnect/it code.