A University of Wisconsin neuroscience researcher falsified “Western blot images as well as quantitative and statistical data” in two NIH-supported papers and three unfunded grant applications, the U.S. Office of Research Integrity (ORI) has found.
As first reported by the Milwaukee Journal-Sentinel and then The Scientist, Rao M. Adibhatla has agreed to retract the two papers, in the Journal of Biological Chemistry and Brain Research:
- Adibhatla, R.M., Hatcher, J.F., Larsen E.C. et al. “CDP-choline Significantly Restores Phoshatidylcholine Levels by Differentially Affecting Phospholipase A2 and CTP:Phosphocholine Cytidylyltransferase after Stroke.” J. Biol. Chem. 281:6718-6725, 2006 (cited 65 times, according to Thomson Scientific’s Web of Knowledge)
- Adibhatla, R.M., & Hatcher, J.F. “Secretory phospholipase A2 IIA is Up-regulated by TNF-[alpha] and IL-1[alpha]/[beta] after Transient Focal Cerebral Ischemia in Rat.” Brain Research 1134:199-205, 2007 (cited 36 times)
committed research misconduct by falsifying Western blot images as well as quantitative and statistical data obtained from purported scans of the films. The research studied the effect of cerebral ischemia on phospholipid homeostasis in an experimental animal model (SHR rat) of stroke during the course of reperfusion of the ischemic cortex. The falsified Western blot images and derivative quantitative data describe changes in levels of sPLA2-IIAA, CCT[alpha], and of PLD2 during reperfusion in the ischemic cortex.
Specifically, the Respondent:
Falsified the Western blot data demonstrating sPLA2 expression in a time course after ischemia in Figure 1B of the JBC paper and Figure 2A and 2C of the Brain Research paper by rearranging the bands such that the labels do not accurately portray what is in the lanes. He perpetuated the falsification by presenting the quantification of the single falsified Western blot in a bar graph as the average of five (5) replicate Western blots. The result in the paper cannot be substantiated by the actual experiments.
Falsified the Western blot data demonstrating CCT[alpha] expression in a time course assay after ischemia in Figure 2A of the JBC paper by rearranging the bands such that the labels do not accurately portray what is in the lanes. He perpetuated the falsification by presenting the quantification of the single falsified Western blot in a bar graph as the average of four (4) replicate Western blots and the six (6) hour time point was further falsified to make the results look better. The result in the paper cannot be substantiated by the actual experiments.
Falsified the quantification of a Western blot demonstrating PLD2 expression in a time course after ischemia in Figure 3A of the JBC paper by claiming a bar graph quantifying a single Western blot is the average of four Western blots.
Submitted the same falsified Western blot images and bar graph data in three unfunded grant applications: NS042008-05, NS042008-05A1, and NS042008-05A2. Specifically:
- the falsified sPLA2-IIA data were submitted as Figures 3, 8, and 12 in the respective NS042008-05, -05A1, and -05A2 applications
- the falsified CCT[alpha] data appeared as Figures 10, 15, and 16 in the respective -05, -05A1, and -05A2 applications
- The falsified PLD2 bar graph data and associated statistical claims appeared as Figures 8 and 13 in the -05 and -05A1 applications respectively.
Adibhatla won’t be eligible for NIH grants for two years, and won’t be able to serve on any NIH committees — including peer review committees — for three years.
We’ve contacted him, along with the University of Wisconsin, and will update with anything we learn.
Update, 2 p.m. Eastern, 1/28/13: The University of Wisconsin tells Retraction Watch that Adibhatla is not tenured, and that his contract will not be renewed when it ends in the next few months:
It is distressing and disappointing whenever we encounter situations where research integrity is compromised, which, thankfully, is not often.
Wow, funding rates are so bad he could not get an NIH grant even after cheating!
What do you know about this PI to comment like this?. Did you ever notice prevention methods failed here?
1. Before publishing the paper reviewers reviews this paper, why they didn’t identify this error?
2. After publishing the paper bunch of scientist’s like you read these papers, why they didn’t identify this error?
According to me if PI identify the error in the paper he should have retracted this paper for corrections, he didn’t identify the error until somebody bring to his attention.
Good point, Real Science. It’s like the guy who uses drugs and STILL loses the race! Pretty bad,
Geez. He fakes NIH-funded data in two journal articles and tries to hoodwink the feds in three more grant applications, and they are making him wait two whole years before he gets to try again. Talk about harsh punishment!
On the bright side, he gets five stars for efficiency. No wasted time, money, or chemicals in his lab. Bet he doesn’t even need much in the way of equipment. So he can propose some economically attractive projects.
I may be wrong, but I think over 24% of all retractions are due to image manipulation so it is arguably the most serious problem we all need to tackle.
Yes, it is something quite doable.
Indeed. Image manipulation is a huge issue, and there is a simple reason why. Image manipuation is done using Photoshop, and the image manipulation is done by interactive programs. This is a situation which is BOUND to fail. If you want REPRODUCIBLE IMAGE MANIPULATION, SCRIPTED methods must be used. I am writing a grant application about this.
Not sure I follow. What situation is bound to fail?
The use of interactive software (such as Photoshop) to process images which are published in journals. That is bound to lead to either incorrectly processed images. Since no reproducible process of image manipulation is used, no one can verify the connection between the original and the final processed image.
Left a phrase out “either incorrectly processed images or fraudulently manipulated images.”
It sounds an interesting idea. I expect you have better concepts on this than me, but could you envisage some kind of Photoshop add-on that institutions could lock on to their copies of Photoshop?
Are you envisaging something that a journal could press replay and see the versions of the image rewind on the screen – perhaps that is not feasible or make files too bulky?
Or something that would just add information to the file’s metadata, eg image cropped [time stamp], image spliced with another image [timestamp], clone stamp applied [coordinates, timestamp]. And this metadata is submitted to the journal?
Speaking of image splice, does there appear to be one in this paper?
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3165912/
Figure 6. I won’t say where precisely, to check that I am not just seeing things
@Littlegreyrabbit, I don’t see a problem with figure 6. There is some pixelation, and you also need to be mindful that images in pdfs are jpegs, which is a lossy algorithm optimized for “typical” photographs (portraits, landscapes), so you can get artifacts from over-compression that have nothing to do with intentional manipulation.
Thanks Strongdreams, I am aware of the issue of pixelation. And I agree that there is the possibility that compression can create an artefactual appearance of a splice line, of course that means provided scientists were careful to match the background exposure they would be free to splice to their hearts content – but this is why we have committees of investigation, to check such issues (I am still unable to spot where and how Maya Saleh’s group manufactured their Nature figure). The deck is already stacked heavily in favor of the fraudsters, and even though image manipulation is going to be least significant cause of misconduct, it is the easiest to detect. It would lovely if everything was as naive Adibhabla who actually spliced his 45 kDa actin strip BENEATH his 14 kDa protein of interest (and even then it is pretty difficult to see for certain the splice lines in Figure 1b)
http://www.jbc.org/content/281/10/6718/F1.expansion.html
Can you say which lane and which strip of Figure 6 you specifically don’t see a splice line – so we can check we are on the same page here.
If you check this figure from the same group you can see very obvious splicing
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2830975/figure/F2/
Upper row of blots.
Some of them are so obviously spliced that its difficult to know if this is fraud or not. But some of the strips don’t appear to be spliced which would mean that the strips weren’t all from the same experiment.
So they do splice and they do know how to adjust backgrounds so they match.
The shocking this is that JBC and Brain Res and really top-tier journals. If such journals of high reputation apparently have no quality control measures in place to verify image manipulation, and these guys have pretty strict peer review, and 99% of the scientific quality has been rigorously verified, then scientists are pretty much all screwed. Photoshop, and even basic Word and Power Point tools, or even Paint, are so often used to change, assemble, etc. Someone needs to publish a paper on guidelines and these need to be published open access for the community to get some advice. There seems to be a disconnect between ORI rules, NIH-funded research grant allocations, QC and plain common sense.
JudyH and S,
Now imagine you being a PI, but doesn’t even know how to run a Western blot, depending on co-authors or postdocs for those results. Please refrain from hurtful remarks because he is one of the most ethical person I ever met. I know him for the past 20 years as a scientist and can’t imagine he knowingly falsifying the data. His reputation and scientific career are crushed for being a PI, actual culprit is free. Eventually it is his fault trusting his team’s ethics and not cross checking the results presented to him. He spent tens of thousands of dollars in lawyers fee to clear his name but didn’t stand a chance. This investigation went on for 4 years, can you imagine the mental torture he has been through?, your sarcasm is not taken well. I feel scientific community lost one of the most sincere and dedicated scientist. He just moved on from science if that makes you feel better, but for me it is a tragedy!
As someone who has supervised students and postdocs doing techniques that are not my area of expertise, I can say that you have to be certain you understand the methods well enough to take full responsibility for them. If not, this is what you get. You are senior author, the buck stops with you. If he did not understand that, then it is best that he is no longer a PI.
Agreed, there seems to be lapse in PI role. I was responding to the jokes cracked at him, knowing his integrity and character I thought it is unfair. Just happens to be standing on the wrong side of issue.
@Reddy: You said RMA is a PI. In the above listed two papers (JBC and Brain Research) he is the first author. In the second paper, there are only two authors – who did the work. These two authors are common in the two papers retracted. Who is the PI and who are his team members. If he has not done those critical experiments – what is his contribution? If he is innocent, who did it…the only other person in the two author paper? Your arguments do not make sense especially these two papers where he is the first author and the corresponding author..
These two publications are result of work done with the support of NIH grant, in which he is a PI. The whole issue is about Western blot which is not done by him. It is not for me to prove who did it, there are several other things like research note books can prove who did what. But he had the responsibility to see how the data is generated, that is where he is paying heavy price. It looks unfair to me because after 30 years of dedicated research, you brand someone cheat and let him lick wounds for rest of his life?. Instead his past record and the fact that the data in question is not generated by him should have been taken in to consideration. The impact of 2 years (he already served one year, that makes 3 years) ban is basically kicking him out of science for ever!
ORI did an investigation, an investigation that you say lasted for four years. Did ORI fail to inspect research notebooks? Did Dr. Adibhatla refuse to give ORI the name of the real culprit? I assume that the investigators interviewed the very short list of co-authors and concluded that Dr. Adibhatla carried out the falsification. If you have evidence that Dr. Adibhatla did not do it, I think ORI will be interested to see your evidence.
Ressci Integrity and nickxdanger make excellent points. Dr. Adibhatla has placed his name first in the list of authors for both papers, indicating that he did the majority of the research. Can he take the lion’s share of the credit when things are rosy yet retreat into the role of uninvolved project supervisor when problems are discovered? How does his ignorance of Western blot technique absolve him of responsibility when the allegation is that the image was manipulated after being captured and that false statements were made about how many replicates are represented?
Can you explain what you mean by your assertion that Dr. Adibhatla “just happens to be standing on the wrong side of [the] issue”? It does not appear to me to describe with accuracy anything about this case.
I do not think it matters if the PI himself falsified the Western blots. He probably created an atmosphere in which doctoring data was both expected and accepted. PIs usually do not do actual research, so there is no reason to expect that they typically falsify data. In a murder-for-hire, a person who pulls the trigger often gets a lighter sentence than the person who ordered the kill. The analogy with scientific fraud should be apparent.
I agree if student is undergraduate, graduate or PHD trainee, not for post doc. Post doc comes with PHD degree and few years or few months experience.
I agree if student is undergraduate, graduate or PHD trainee, not for post doc. Post doc comes with PHD degree and few years or few months experience.
Specifically who are you talking about? Is the “ethical but ignorant PI” the subject of this RW piece, or a different person? I am not asking you to identify them, just to indicate if this person is the subject of this discussion.
It’s about Adibhatla, who ever interacted with him will be stunned to see this development.
His ability to do Western blots is in fact totally irrelevant. It’s the processing of the images which is the issue. As they note above in the text, the bands were re-arranged. The technician who performed the WBs may not be the person who re-arranged the bands. The PI, however, is responsible for the manner in which images go from the raw form to the form published. That is the moment of fraud.
I wonder where he will end up. He is not going to be continued at UW.
Yes, processing of images is the issue. Single band was duplicated to show the protein levels didn,t change over a period of time. Investigators did check for the raw data, surprise! lot of data related to those time points on Western blot weren’t recorded in the notebooks by the person conducted those experiments. This is the clue investigators (consisting 2 scientists and one lawyer) are ignoring which is pointing to “the moment of fraud”. Don’t come back saying the PI did something to raw data because the investigators seized all notebooks, computers all relevant things before they even notified him of the fraud investigation.
If the PI created the Western blot without data provided by the coauthor, coauthor would have alerted concerned authorities in 2007 itself. On what basis you say “The technician who performed the WBs may not be the person who re-arranged the bands”?. If he is too lazy, copy and paste is easier than conducting experiments, then why not the person performed the WBs duplicated the bands? For me only he can duplicate because there is no one to complain about data integrity. This PI is just interested in results.
To your other question “where he will end up”, he left science for good.
JudyH,
Glad to see your posting on a serious note. As far as I know ORI went with the university investigation, didn’t conduct its own investigation. Hope in my reply to nickxdanger I answered some of your questions. Investigators had all the evidence in front of them but I feel they selectively ignored some of the facts and made him a scape-goat.
As a first author and PI he did take responsibility and left science. The sarcastic BS written earlier by some dragged me in to this discussion.
It took 4 years, costed Adibhatla ~$40,000 in lawyer’s fee and he doesn’t have any more money to defend himself. It is not possible to answer all your questions in few replies. Though it doesn’t change the outcome, if you are in and around Madison, WI I will be glad to have a thorough discussion with you in person.
In two years, his collaborators and post-docs will be long gone as well as his post at Wisconsin.
N&S
Please Stop wondering about him where he will be, he is a great scientist it is not my word it came from bunch of people from his department, he learned his lesson hard way trusting others, definitely his knowledge put him in the top position in some other field.
Who are these mysteriously “others” whom he trusted? I’m sure ORI would like to know.
Perhaps the people he bragged about regarding making up stuff?
There are several papers that should be retracted from this author. Didn’t ORI check all his papers? Check out Figure 8 in this paper for example:
http://www.neurosurg.wisc.edu/labs/adibhatla/pdf/2007_Neurosci_PC12.pdf
Junk Science
http://www.neurosurg.wisc.edu/labs/adibhatla/pdf/2007_Neurosci_PC12.pdf
I always tell students one thing about quantitation of blots:
“You cannot quantitate a saturated band”
You mention figure 8 – all but one of the B-actin bands are saturated. In my experience it is impossible to quantitate such bands.
Figure 4 A and B – look at the squashed B-actin bands in A and compare to the saturated fattened B-actin bands in B.
I agree, it should be in the author guidelines for all journals. Abstract 200 words and no saturated bands etc etc. The cutting and pasting in Figure 8 is the worst photoshopping job I’ve seen in a while.
Sorry to ask a stupid question but how can you tell in the lab or in the images that a band is saturated?
No such thing as a stupid question…
One way could be to compare the developed film to a strip of other film which has been exposed to full light (which would be just black) – if any of your bands on the developed film are just black and without any grey tones, you are probably looking at a saturated image.
Some imaging systems have software programs that will highlight saturated bands when the image is being visualized. Another approach would be to run a “standard curve” gel with increasing amounts of protein loaded…i.e 5, 10, 20, 40, 80 micrograms etc. Ideally when running your samples of interest you would load an amount of protein that was in the linear range of your standard curve. The issue with this is that often your protein of interest and house keeping protein are expressed at different levels and it can be difficult to avoid saturated house keeping proteins.
X-ray film only has so much silver that can be exposed. The best way (IMO) to quantitate Westerns is to make multiple exposures and scan them all and graph the intensities to make sure they are linear. If the graph tops out you have a saturated band and need to go back and pick a shorter exposure. It is very easy with control bands (GAPDH, tubulin, actin) to use the wrong dilution and get saturated bands with just a few second exposure because students these days are not taught proper theory. That said, some direct imaging systems count photons and can’t really saturate, but might give you a printout that looks saturated, so you have to consider the actual method used, not just the appearance.
thanks most informative…..any good references you can suggest for ideal Western Blot technique and analysis>
Thanks in advance from an interested clinical researcher
It’s not just that the bands are oversaturated. Zoom in and look at the top and bottom of the PARP blot. I can clearly see that lane 3 (OGD 0.44) overhangs on the bottom, lane 4 (OGD 0.54) overhangs at the top, lane 5 is not lined up, and lane 6 doesn’t line up. I’m normally terrible at the “spot the problem” game, but that one (Figure 8) is pretty obvious.