The U.S. Office of Research Integrity has sanctioned Bryan William Doreian, a former postdoc in dermatology at Case Western Reserve University in Cleveland, for falsifying data in his dissertation and a 2009 paper in Molecular Biology of the Cell (for which it provided a cover image, at right).
ORI says Doreian’s bad NIH-funded data also appeared in a manuscript submitted to, but never published in, Nature Medicine.
Here are the papers cited in the finding:
- Doreian, B.W. “Molecular Regulation of the Exocytic Mode in Adrenal Chromaffin Cells.” Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy, August 2009; hereafter referred to as the “Dissertation.”
- Doreian, B.W., Fulop, T.G., Meklemburg, R.L., Smith, C.B. “Cortical F-actin, the exocytic mode, and neuropeptide release in mouse chromaffin cells is regulated by myristoylated alanine-rich C-kinase substrate and myosin II.” Mol Biol Cell. 20(13):3142-54, 2009 Jul. (cited 14 times, according to Thomson Scientific’s Web of Knowledge)
- Doreian, B.W., Rosenjack, J., Galle, P.S., Hansen, M.B., Cathcart, M.K., Silverstein, R.L., McCormick, T.S., Cooper, K.D., Lu, K.Q. “Hyper-inflammation and tissue destruction mediated by PPAR-γ activation of macrophages in IL-6 deficiency.” Manuscript prepared for submission to Nature Medicine; hereafter referred to as the “Nature Medicine manuscript.”
As a result of the Respondent’s admission, the Respondent will request that the following paper be retracted: Mol Biol Cell. 20(13):3142-54, 2009 Jul.ORI finds that Respondent falsified numerical values in the Mol Biol Cell paper, the submitted Nature Medicine manuscript, and the Dissertation by altering the number of samples or the experimental results to improve the statistical results. Specifically, ORI finds that Respondent:
-
falsified the quantification of immunofluorescence for the ratio of phosphorylated to unphosphorylated MARCKS protein in response to different stimuli in Figure 2 of the Mol Biol Cell paper and in Figure 12 of the Dissertation by falsifying the sample number as n=15
-
falsified the quantification of immunofluorescence for filamentous actin in response to different stimuli in Figure 3 of the Mol Biol Cell paper and in Figure 13 of the Dissertation by falsifying the sample number as n=15
-
falsified the quantification for the effect of blebbistatin on catecholamine release as determined by patch clamp analysis in Figure 22 of the Dissertation by stating that 14 cells had been assayed when only 8 cells had been assayed
-
falsified the Pearson’s cross-correlation analysis in Figure 7 of the Mol Biol Cell paper and in Figure 25 of the Dissertation, used to calculate the degree of spatial correlation between pan-chromogranin A/B (CgA/B) and the endosomal membrane, by stating that 20 or more cells had been tested for each condition when only 9-18 cells had been tested for each condition
-
falsified RT-PCR values for iNOS and TNF-alpha expression recorded on spreadsheets and presented in Figures 5e and 5f of the Nature Medicine manuscript showing the effect of hyper-inflammatory macrophage generation on tissue destruction, by falsifying the numeric values to fit the hypothesis of the manuscript
-
falsified ELISA graphs for the concentration of TNF-α in the aAB IL-6 mice and their controls in Figure 6j of the Nature Medicine manuscript showing the effect of rosiglitazone treatment in the mice, by multiplying the experimental values by 100 to match the magnitude of the values presented in Figures 21, 6h, and 6i of the Nature Medicine manuscript
-
falsified the RT-PCR results presented in the Nature Medicine manuscript for quantification of iNOS and TNF-α RNA expression by claiming that the results represent the rmean of three identical experiments when the three experiments were normalized differently to yield the desired result. Specifically, false results were presented for peritoneal macrophages treated in vivo with rosiglitazone and/or inhibitors of PPARγ signaling Figures 1g, 1h, and 1i, and for iNOS RNA expresssion in IL6-/- macrophages treated in vitro with either SOCS3 antisense oligonucleotides in Figure 2g or the STAT3 decoy in Figure 2j.
Nothing on the Mol Biol Cell website indicates that Doreian’s article has been retracted, but we’re guessing that will change shortly.
Doreian also has a 2008 paper in the Journal of Neuroscience, “Myosin II Activation and Actin Reorganization Regulate the Mode of Quantal Exocytosis in Mouse Adrenal Chromaffin Cells,” that isn’t mentioned in the ORI report and does not appear to be tainted. It has been cited 35 times.
Doreian has agreed to a three-year settlement in which his research activities will be supervised. That may be moot, however. Doreian seems to have created a career outside the lab as the head of a firm that helps patent attorneys pass the bar.
“Apart from that some people say I am pretty good at data analysis and organization.” – it probably depends on the definition of data analysis.
Yup. *snickers* According to his timeline, he left science in the summer of 2012 (probably knowing that his days were numbered) and turned his attention to the patent bar exam. Hm. I doesn’t say he had always wanted to be a lawyer. But anyway. So he did a phenomenally quick analysis of the bar exam and, despite having taken no classes at a law school, was able to pierce the dark shroud surrounding the exam and illuminate the basic approach needed to clear the bar (pun intended). I wonder if his insight into passing the bar exam is anything like his insight into producing impressive results in the lab. He didn’t himself pass the bar exam, but he’s teaching other people how to do it. Right.
Obvious concern: the PhD thesis containing the raw material for the bogus paper is not retracted?
I don’t think ORI has the authority to require a university to withdraw a thesis, and thereby withdraw a degree. Probably this is referred to the Dean or Provost or someone at the university for consideration.
What is interesting to me is that most of the errors (except #5, and maybe #7 which is not clear) are cases of taking data showing an effect, and making it look better by inflating the N. ORI does not say the findings themselves were false, only that he claimed more replicates than he actually performed. I wonder if the data could have been published if he had simply offered the correct, lower N. It may be possible to correct the thesis, assuming Dr. D. is still interested in keeping his degree.
Yes, it would be interesting to find out whether he reported a higher N because somebody told him he needed that many or because he couldn’t get a statistically significant result using a lower N.
Number 6 says he multiplied his actual results by 100 for both the experimental units and the controls in the Nature Medicine manuscript. Could that have been an honest error? Failure to understand how to do the calculation, but doing it the same for all mice?
On multiplying; maybe the lower results were equally valid but different magnitudes because of some experimental difference (different cell type, different stimulus, etc) and he just goosed it to make it look the same to avoid questions. Or it could be that the experiment with the low values had a technical fault and he should have replicated it, but was lazy. (That seems to be the best explanation for inflating the Ns.) We’ll likely never know.
I’m just fascinated (and slightly horrified) that people are losing their careers because it is easier to photoshop a western blot that looks good than to just re-run it clean (which is what I think many of these cases are).
This looks like someone from the inside dobbing him in, possibly the head of his laboratory, as there is nothing that looks visible to the reader. In which case I assume they had good reason for doing so and the ORI have just selected those issues in which the case can be proved beyond any dispute.
“I’m just fascinated (and slightly horrified) that people are losing their careers because it is easier to photoshop a western blot”
Who has lost their career for simply photoshopping a western blot? Most people don’t lose their careers. As far as I am aware anyone who has come unstuck with a western blot has presented data claiming quantification from loading controls that did not belong to that experiment.
On the other hand with this paper
Vaccinia virus protein F1L is a caspase-9 inhibitor.
Zhai D, Yu E, Jin C, Welsh K, Shiau CW, Chen L, Salvesen GS, Liddington R, Reed JC.
In figure 2b
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2820784/figure/F2/
They have sliced together two western blots at the lane after the “smile”, the upturned bands. In some of the strips you can spot the splice lines, in some you can’t, if you didn’t have the guidance of the upturned band you might lack confidence to say this had been spliced at all. It appears to be a group that has acquired considerable (and one hopes unnecessary) expertise in the dark arts of photoshop to join blots together.
In this case there is no reasonable scenario that would suggest they were deceiving in regards to what these blots show. They just spliced them together because they can. This group appears to do this quite a bit as far as I can tell.
Is it too much to ask for that they should be encouraged to desist from this practice?
“Who has lost their career for simply photoshopping a western blot?”
There certainly are people who seem to be losing their careers over photoshopped western blots. I guess without studying them in depth I don’t know how many cases are making results prettier (that are probably real), and how many cases are making results up out of whole cloth.
“Is it too much to ask for that they should be encouraged to desist from this practice?”
I’m an academic editor for one of the plos journals, I certainly look at these things a lot more closely now. If there is a splice it needs to be clearly marked and explained in the legend (such as; 2 ten-well gels showing the results of an experiment with more than 10 conditions).
PLoS ONe has many many academic editors as well as editorial board members. This might be a problem – not sure how they are chosen!!!
Perhaps if you present what you think is the most egregious example – generally you find others things were going on besides just a desire to tidy up a blot.
As an advising editor can you tell me what is recommended practice in this situation when probing for proteins and their phospho equivalents
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1851032/figure/F5/
Proc Natl Acad Sci U S A. 2007 Apr 10;104(15):6371-6. Epub 2007 Apr 2.
Ubiquitin-conjugating enzyme Ubc13 is a critical component of TNF receptor-associated factor (TRAF)-mediated inflammatory responses.
Fukushima T, Matsuzawa S, Kress CL, Bruey JM, Krajewska M, Lefebvre S, Zapata JM, Ronai Z, Reed JC.
It seems to me that many of the strips don’t belong to the same gel. For instance Figure 5b, p38 and p-p38 – one has a smile, the other doesn’t. Ditto Figure 5e, jnk and p-jnk. I thought the practice was to strip the blots and reprobe them Or is it normal just to run two gels in parallel, loading equals amounts on both?
In reply to littlegreyrabbit:
this figure looks like they pasted two gels with 10 lanes each (standard format) together, if all is well, from the same fractionation experiment. The clean way to do this, is to put a thin line between the two parts to indicate that this is a composite image. That doesn’t diminish the quality of the data or the presentation.
While I don’t think there is any possible nefarious intent beneath that particular figure, I was just using it for an illustration how much work and how mannered their representations of westerns are
However, this paper
Proc Natl Acad Sci U S A. 2005 October 18; 102(42): 14982–14987.
Published online 2005 October 11. doi: 10.1073/pnas.0507512102PMCID: PMC1257734Applied Biological Sciences
Method for targeting protein destruction by using a ubiquitin-independent, proteasome-mediated degradation pathway
Shu-ichi Matsuzawa, Michael Cuddy, Toru Fukushima, and John C. Reed*
In particular figure 3
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1257734/figure/fig3/
Click on the image to the fullest extent that PMC edition allows, you can see all the westerns have been deposited onto a gray background – sometimes you can see the join marks of the original membrane against this background, but sometimes the membrane background has been adjusted so it merges seamlessly into this artificial grey base.
Why do this?
I have annotated where it would have been easily possible to have manipulated and spliced this image where it would not be detectable due to this very mannered method of presenting their blots
http://img138.imageshack.us/img138/1559/tempxw.jpg
I have not amplified the images above what PMC provided – obviously it is getting a bit pixelated.
I can’t say that these images have been manipulated, all I am saying they could have been without leaving any indications that the eye could see.
Most people just present their blots without such extensively working on them, in a way you can be absolutely sure that nothing untoward could go on. These guys? Who knows.
There’s a chance that the uniform background is an artifact of multiple rounds of compression. But if I saw this as an editor or reviewer, I would ask to see the uncut, unaltered blots.
Yes LGR, I see what you’re getting at. Similar things can be seen in this paper from the same group:
http://www.jbc.org/content/early/2007/11/21/jbc.M708419200.full.pdf+html?with-ds=yes
For example, in Fig. 1B lane 1-173 looks like a different exposure of the 35S-INT 50% lane, and the 1-99 lane looks the same as the 35S-INT 50% lane, but it is hard to tell coincidence from duplication. It looks like the 1-81 band has been truncated, and in Fig. 1C it looks like there is a gray rectangle pasted at the bottom of the FL lane, but it is hard to be sure. In Fig. 2C the FL lane seems to be truncated horizontally, Fig. 4A the last lane looks like it’s pasted in; Fig. 4B the first lane (anti-HA) seems to be missing; Fig. 8, 9 the contrast is very high and the gray background looks so uniform it seems artificial.
John Reed has 11 R01s. You can’t expect him to look at every piece of his lab’s data!!!
http://www.nature.com/news/2008/080319/full/452258a/box/1.html
Is this person famous for apoptosis research? A big name in the field, I guess.
I question the wisdom of study sections that award 11 R01s to a single investigator.
And now he is founder of Wysebridge, a company that sells patent law exam materials…. He screwed a lot of people over, including other students who tried to repeat his work or use his data as a platform for their projects that never got off the ground. He also claims “discovery and development of a hyper-inflammatory phenotype of macrophages generated through activation of PPAR-gamma receptors, to study delayed and chronic wound healing conditions” on http://bryandoreian.brandyourself.com/. He neither discovered nor developed, he came in on a project already a year in the making by the head of the lab and his research assistant….
here’s the list of funders of this fraudulent research according to the federal register. anybody besides me have family members who could have been helped by breakthroughs in treatment/cures for these illnesses?
” research supported by National Heart, Lung, and Blood Institute (NHLBI), National Institutes of Health (NIH), grant T32 HL07887 and National Institute of Neurological Disorders and Stroke (NINDS), NIH, grant R01 NS052123.”
Any news on the retraction?