Retraction Watch

Tracking retractions as a window into the scientific process

Paper by Canada Research Chair retracted from journal he edits for blots “from unrelated samples”

with 16 comments

skeletal muscleA lab run by a Canada Research Chair at the Ottawa Research Institute has retracted a paper — in a journal the chair edits — for what sounds a lot like inappropriate image manipulation.

Here’s the notice from Skeletal Muscle:

The authors would like to retract the article “MyoD-dependent regulation of NF-κB activity couples cell-cycle withdrawal to myogenic differentiation” [1]. After the article was published, the first author revealed that the tubulin blots in Figures 1A, 3E, and 4A are from unrelated samples. The lead author and co-authors apologise to the readers, reviewers and editors for publishing this erroneous data.

The first author, Maura Parker, is now at the Fred Hutchinson Cancer Research Center. Neither she nor corresponding author Michael Rudnicki — who “holds the Canada Research Chair in Molecular Genetics and is an International Research Scholar of the Howard Hughes Medical Institute” — have responded to our requests for comment. Rudnicki is also one of the three editors in chief of Skeletal Muscle. (Rudnicki’s auto-responder said he was unavailable until August 8.)

The work was supported by Rudnicki’s grants

from the Canadian Institutes of Health Research, the Muscular Dystrophy Association, the National Institutes of Health, and the Canada Research Chair Program, and by grants to [author Denis Guttridge of Ohio State] from the National Institutes of Health.

We’re not sure what prompted the retraction, but we do know that the pseudonymous Clare Francis contacted Ottawa officials in April with suspicions of data fabrication.

Written by Ivan Oransky

July 24th, 2013 at 9:57 am

Comments
  • Toby July 24, 2013 at 11:51 am

    Fig 6D also looks like a jigsaw puzzle!

  • fernando pessoa July 24, 2013 at 12:03 pm
    • fernando pessoa July 28, 2013 at 6:14 am

      Is the stated reason for the retraction “tubulin blots in Figures 1A, 3E, and 4A are from unrelated samples”
      only part of the reason?

      Figure 4A, WT (left) tubulin panel. Light area at the end of the band in lane 2 and a step in the level of the bands between lanes 2 and 3. Right tubulin( MyoD-/-) panel change in background between lanes 3 and 4. These would fit the stated reason.

      Figure 4A . Right cyclin A (MyoD -/-)panel. Bands right 2 lanes very similar.
      Right cdk2 (MyoD-/-) panel. Horizontal changes in background in right-most lane.
      Left cyclin E (WT) panel. Bands right 2 lanes identical.
      The latter does not fit the stated reason.

      Please compare the MyoD-/- panel of figure 4A with the MyoD-/- cylin D3 panel of figure 3E.

      http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3356597/figure/F3/

  • Deidentified July 24, 2013 at 12:04 pm

    Reblogged at the…just kidding!

    That said, I shudder in fear at the day someone designs a little program that generates nondescript images of blots that can be turned into a figure as desired.

    Even new versions of Agilent’s Chemstation come with a bewildering array of GLP options and chain-of-custody setups that track almost every change done to the digital copy. That said, academic research has operated as a system of peers on trust; and I worry that those days are fast fading away…

    • AGMMGA July 26, 2013 at 10:32 am

      Who needs software? Western blot is already the easiest technique to manipulate. Forget image manipulation, simply fake your gels!
      Want a perfect invariant control? just load the same sample n times. Want a nice time-dependent increase / decrease? load the same sample in a serial dilution, ascending or descending as desired. Want a blank lane? Load nothing.Also, forget about weird uncharacterized antibodies… all bands are black, so just use GST (or whatever you have at hand), then label it whatever you want.
      Seriously, if you are out for fraud, western blot is your friend. I guess things could improve if editors and journals demanded pictures of full gels instead of allowing the ridiculous collage work that is common these days, but I am not holding my breath that this is gonna happen or that it is gonna make a difference anyway…

      • fernando pessoa July 26, 2013 at 11:12 am

        AGMMGA July 26, 2013 at 10:32 am

        “I guess things could improve if editors and journals demanded pictures of full gels instead of allowing the ridiculous collage work that is common these days”.

        Many think that the ridiculous collage work is cool.
        They have their model with arrows showing what should go up or down, and by kicking the cells hard enough they can get some results to fulfill expectations, but often not all at the same time. That’s where the collage work comes in so handy. It gets you out of tricky situations and can go on forever. You simply put the westerns in a drawer and hunt through them for something that looks like what you want. You can even get fond of them and use them again. Stranger things have happened.

        http://web.archive.org/web/20130415035452/http://www.science-fraud.org/wp-content/uploads/2012/08/SWLee17.jpg

        It might actually help if somebody, it might be the editors, or the people who tidy up the office at the end of the day, looked at what was presented for publication. Asking for pictures of full gels is idealistic, starting with small steps has a higher chance of happening.

        There are 79 publications for the journal Skelet Muscle in pubmed. They are all free to look at.
        I takes an hour to go through the westerns in them (they do not all have westerns). I thought 7 out of the 79 total publications have odd images. One of these has been retracted (the topic of this post). Some more might need to go. There is a spectrum. They do need to be looked at.

        • fernando pessoa July 26, 2013 at 11:14 am

          The references for the webpage.
          Mol Cell Biol 2000;20;7450-9; PMID11003642
          EMBO J 2003;22;1289-301; PMID12628922

        • AGMMGA July 27, 2013 at 5:01 am

          Good points Fernando.

          Also, asking for full gels only means I have to find multiple “standards” at different molecular weights to fake the gels… or look harder in my drawers for a good blot to paste in. Still not impossible.

          I suppose science will always be based on an implicit bond of trust – as a reviewer, I’ll check you have not made mistakes, but I cannot quite check that you have not faked your data – unless you are an idiot who sucks at photoshop. To see a silver lining in these retraction clouds, it seems that a lot of fakers are actually also terrible at fakery.

      • Sham July 26, 2013 at 1:22 pm

        100% agreed..

      • ferniglab July 27, 2013 at 10:28 am

        One might think that in labs that manipulate images, they believe it is Ok to do this, but not to actually fake the gels? This would be quite weird and perhaps the correct deduction is that they are both so lazy and incompetent that they cannot get off their backsides to run another gel and they wouldn’t know how to calculate the loading or spiking to get the gel they want?
        One would need to have the whole Western and a protein stained gel or blot. The sort of things you do anyway as basic quality control, but in some places there isn’t much quality!

  • Junk Science July 24, 2013 at 2:40 pm

    Why was not the other Skeletal Muscle paper retracted at the same time, look at Fig. 3A:
    http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3299591/figure/F3/
    Some splicing in the Mol Cell Biol paper as well.

    • fernando pessoa July 24, 2013 at 5:11 pm

      Figure 3F. J Cell Biol. 2008 Feb 25;180(4):787-802.
      http://jcb.rupress.org/content/180/4/787/F3.large.jpg

      Primary myoblasts and C2C12 myoblasts p65 panels. The alpha-Tubulin panels are different.

      • Junk Science July 24, 2013 at 6:33 pm

        Yep, those p65 blots looks identical. I am tempted to say that the 4 and 24 weeks of Fig. 2B are consecutive sections of the same sample, because it seems highly unlikely that the cell pattern are that similar.
        http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2536604/figure/F2/

        • fernando pessoa July 27, 2013 at 8:04 am

          In reply to Junk Science July 24, 2013 at 6:33 pm

          Mol Ther. 2008 July; 16(7): 1340–1346.

          I am impressed by your observation and inference that “the 4 and 24 weeks of Fig. 2B are consecutive sections of the same sample, because it seems highly unlikely that the cell pattern are that similar”.
          I think that you are more than likely correct for the very reason you give.

          The penultimate author does have this pubpeer entry

          http://pubpeer.com/publications/15466486

  • james December 30, 2013 at 4:27 pm

    Michael Rudnicki has just been named to the Order of Canada.

    “The Order recognizes outstanding achievement, dedication to the community and service to to nation.”

    Congratulations to Dr. Rudnicki?

    • david hardman February 19, 2014 at 12:54 pm

      2014 correction of 2004 paper posted as a comment by the senior author.
      Will that appear in the Pubmed record?

      http://www.plosbiology.org/annotation/listThread.action?root=74973

      Original article.

      PLoS Biol. 2004 May;2(5):E130. Epub 2004 May 11.
      Pax7 is necessary and sufficient for the myogenic specification of CD45+:Sca1+ stem cells from injured muscle.
      Seale P, Ishibashi J, Scimè A, Rudnicki MA.
      Author information

      Department of Biology, McMaster University, Hamilton, Ontario, Canada.

      PMID: 15138500

  • Post a comment

    Threaded commenting powered by interconnect/it code.