Advertisement

Crystallographer faked data

A protein researcher at the University of Alabama at Birmingham (UAB) has been found guilty of falsifying data that he used to construct 12 fraudulent protein structures that made it into the scientific literature and an international archive of protein structures.A G-protein image based oncrystal structure dataImage: S. Jahnichen After investigating the misconduct -- with the help of a committee of independent protein scientists -- the university has linkurl:asked;http://main.uab.edu/Sites/rep

By | December 18, 2009

A protein researcher at the University of Alabama at Birmingham (UAB) has been found guilty of falsifying data that he used to construct 12 fraudulent protein structures that made it into the scientific literature and an international archive of protein structures.
A G-protein image based on
crystal structure data

Image: S. Jahnichen
After investigating the misconduct -- with the help of a committee of independent protein scientists -- the university has linkurl:asked;http://main.uab.edu/Sites/reporter/articles/71570/ that the structures be removed from the database and that ten research papers, authored by former UAB researcher H.M. Krishna Murthy over the past decade, be retracted from the literature. "What we know is that when Dr. Murthy was asked to provide the data behind the structures, there was not sufficient material presented to allow the expert panel to determine the source of the error," linkurl:Richard Marchase,;http://main.uab.edu/show.asp?durki=8039 UAB's vice president for research and economic development, told __The Scientist__. The UAB investigation concluded that the structures Murthy proposed violated basic physical and chemical laws, making their existence virtually impossible. "There were just many aspects of the proposed structures that didn't appear to be at all plausible given the physical laws of how proteins come together," added Marchase, who is also UAB's scientific integrity officer. "I think [Murthy] deserves tarring and feathering," linkurl:Gert Vriend,;http://swift.cmbi.ru.nl/gv/start/index.html a bioinformatician at Radboud University Nijmegen Medical Centre in The Netherlands, told __The Scientist__. Murthy, whose UAB contract at the school's Center for Biophysical Sciences and Engineering expired in February of this year with the university opting not to renew it, "has always denied any misconduct," according to Marchase. Murthy could not be reached for comment by email, and UAB is unaware of his current whereabouts. According to Marchase, the university informed the Office of Research Integrity (ORI) of the U.S. Department of Health & Human Services that UAB was investigating Murthy's suspect structures, and presumably the ORI will now initiate their own investigation of Murthy. But the agency is not talking about the Murthy case yet. "We can neither confirm nor deny that we have such a case," wrote an ORI spokesperson in an email to __The Scientist__. Murthy's fraud was uncovered when European crystallographers who were working on some of the same structures that Murthy was claiming to solve raised red flags. linkurl:Piet Gros,;http://www.crystal.chem.uu.nl/group-gros/ a crystallographer at Utrecht University in The Netherlands, was alerted to potential problems with Murthy's structures by one of his grad students. Gros's lab published a structure for the immune protein C3b in a 2006 linkurl:issue;http://www.nature.com/nature/journal/v444/n7116/index.html of __Nature__. Along with the Gros group's paper, two other papers on the structure of C3b were published in the issue. One was from Murthy's group at UAB and the other from researchers at the biotech company Genentech. "We noticed quite quickly that there was a problem with the crystal lattice in the Murthy structure," Gros told __The Scientist__. "There were huge gaps between layers of molecules. That was a huge red flag." Gros said that he contacted Murthy about the abnormalities, but the UAB researcher brushed aside his concerns. "Basically we were not satisfied with his answers to the questions," Gros said. "He could not give an explanation to us." Gros and his colleagues began looking into more of Murthy's structures and noticing other "abnormal features." After unsuccessfully trying to get answers from Murthy, they had independent colleagues, including Vriend, analyze the structures and wrote a linkurl:letter;http://www.nature.com/nature/journal/v448/n7154/full/nature06102.html to __Nature__ and to UAB detailing their findings. Using sophisticated validation software, Gros and his colleagues uncovered several problems with more of Murthy structures. There were unexplained gaps in the packing pattern of amino acids in the structures and contacts between atoms that were unrealistic, and "there was no solvent in the data," according to Gros. "There are things which are physically impossible," he said. Though the problems with Murthy's data seem glaring, uncovering the fraud required intricate analysis by protein experts. "The structure [of C3b] was strange, but it did require an expert to read from the numbers that something was worrisome," said Vriend. Murthy was able to publish fraudulent data on 12 protein structure over the span of 10 years in several different journals, such as __Nature__, __Proceedings of the National Academy of Science__, __Biochemistry__, and __Cell__. At the beginning of this month, the __Journal of Biological Chemistry__ (__JBC__) retracted one linkurl:paper;http://www.jbc.org/content/284/49/34468.full?sid=ac0b26d9-bcad-4e65-9c8a-759156f52433 that Murthy published in 1999 on the structure of a domain in the dengue virus. According to Marchase, "Another journal is likely to retract but has not yet," but he declined to say which one. Officials at the linkurl:Protein Data Bank;http://www.rcsb.org/pdb/home/home.do (PDB), which archives thousands of protein crystal structures from the literature, have removed the structure (dengue virus NS3 serine protease) reported in the __JBC__ paper from its database, and "will make the remaining 11 entries obsolete if and when the corresponding papers are retracted," according to a PDB statement. Vriend and Gros said that one good thing to come out of the Murthy fraud case is that journals and the protein structure archives are rethinking the process whereby they validate submitted crystal structures. "This appears to be an extremely rare case," Gros said. "But the [validation] process has to be improved. People are discussing processes of doing this better, and I think that will happen, and it's a good thing." Vriend agreed, adding that the analysis of Murthy's fraud demonstrated the importance of carefully validating proposed crystal structures for proteins. "Protein structures are so important for so many fields in science that they must be right." __Correction (18th December): In the original version of this story, Murthy was said to have been the recipient of five NIH grants this year. The NIH's linkurl:Research Portfolio Online Reporting Tool,;http://projectreporter.nih.gov/reporter.cfm, inaccurately lists the UAB email and name of H.M. Krishna Murthy under grants won by Krishna K. Murthy of the Southwest Foundation for Biomedical Research in Texas. Krishna K. Murthy was not involved in any way with the fraud perpetrated by H.M Krishna Murthy, and the subject of this story was not awarded any NIH grants in 2009. __The Scientist__ regrets the error.__
**__Related stories:__***linkurl:MRI homes in on protein structure;http://www.the-scientist.com/blog/display/55332/
[12th January 2009]*linkurl:Structure hints at Ebola's cunning;http://www.the-scientist.com/blog/display/54827/
[9th July 2008]*linkurl:Former UPenn postdoc faked images;http://www.the-scientist.com/news/display/53469/
[7th August 2007]

Comments

Avatar of: eve barak

eve barak

Posts: 85

December 18, 2009

\nWow. Think of the implications!\n\nCertainly Dr. Murthy was not in the lab all by himself during those ten years. What in the world was going on there? And how come nobody at UAB noticed?\n\nOnce upon a time, unfortunately a long long time ago, UAB faculty members were (legitimately) at the intellectual forefront of protein structural chemistry. \n\nHow could this have happened, and over such a sustained period of time? \n\nIn my opinion, this situation is a formidable indictment of our current peer review system, which seems to be failing miserably, in terms of both journal review and grant application review. This is exactly the kind of thing that can be expected when reviewers, in awe of a scientist's prior publications (his "track record"!), don't bother to actually review the newly-submitted work (let alone prior publications) for plausibility, but instead just mindlessly praise the researchers because of name recognition. The obvious -- and terrifying -- result is that the next generation of reviewers will be trained in the laboratories of mentors like Dr. Murthy. \n\nOur country is falling behind the rest of the world in science, and while this specific kind of failure of peer review is certainly not the only cause of this decline, it is certainly a contributing factor. Our country NEEDS better than this. \n\nAfter dithering about posting anonymously or not, I've decided to let my name show. Those who know me understand the sincerity and passion that underly my concerns for the future of science in the USA.
Avatar of: anonymous poster

anonymous poster

Posts: 1

December 18, 2009

One thing that the author did not comment on is whether there were massive differences in the structures as well. I understand the problems with unrealistic crystal lattice etc, but how different were the structures? is it something that should have raised red flag right away?
Avatar of: anonymous poster

anonymous poster

Posts: 2

December 18, 2009

While this is a very unfortunate situation you have to ask yourself how many institutions would initiate an investigation of someone bringing in grant money. I can tell you..not many. If you read the article you notice that UAB chose not to renew this PI's contract in February of this year..meaning almost one year ago.\nThis is not an isolated situation among both academic institutions and industry. The University of Alabama at Birmingham (UAB) still serves as one of the "gold standards" of research and medicine, both nationally and internationally.
Avatar of: Bob Grant

Bob Grant

Posts: 22

December 18, 2009

Dear Anon,\n\nFully understanding the mathematics and modeling that underlie the construction of these theoretical protein structures is far beyond my expertise. But crystallographers and bioinformaticians led me to believe that the discrepancies in Murthy's structures were subtle enough to evade detection in peer review, though unusual enough to raise suspicions among some experts. These suspicions were only confirmed after rigorous analysis using sophisticated validation software.\n\nFor more detailed information on what exactly was amiss with Murthy's structures please see the UAB press release detailing the misconduct ( http://main.uab.edu/Sites/reporter/articles/71570/ ).\n\nSome of the problems with Murthy's structures involved chemical components being askew by fractions of an angstrom - not exactly the type of stuff that might raise red flags out of hand.\n\nThanks for reading,\n\nBob Grant\nAssociate Editor, The Scientist
Avatar of: Mitchell Wachtel

Mitchell Wachtel

Posts: 30

December 18, 2009

If this work is indeed proven fraudulent, the question would be how one could think of getting away with it. Those most likely to read the paper with care are precisely those who will test the model for consistency with known data. In studies with humans, variation in results is almost required; not so with structural biology.
Avatar of: Ellen Hunt

Ellen Hunt

Posts: 199

December 18, 2009

I read the link to the assessment. It is not correct to say that the assessment depends on things being off by fractions of angstroms, though in some cases that may be technically true. One angstrom is the nominal size of a hydrogen atom in its ground state. So one tenth angstrom for something on that scale is quite meaningful. Translated to human scale, it would be the length of your hand. \n\nThink of how much difference the length of your hand would make in a diagram of an automobile engine. But if you read the assessment you will see that it appears Murthy just fiddled with copies of other people's structures. Cheating paid for him for a long time. \n\nIn immunology it is far easier to fake data. One can make it up out of whole cloth without too much trouble. Fake gels are easy. Fake flow cytometry data is easier. Faking images isn't all that hard, or stain for one thing and say it's another. I know it happens pretty frequently in two ways. \n\nThe Brazen B@$tard: Make up it up and publish fake new results to agree with some hypothesis. \n\nThe Sneaking Snake: Find some hapless lab-rat who has labored for years to produce data that shows something very interesting. Get hold of the data. (Photograph it with your cell phone, a mini camera, or just take excellent notes, or steal the data off the hapless one's computer.) Then make up your own data that duplicates the real results. Rush it into print. Let the hapless one drown while you get the grant to follow it up. (Repeat until tenured chair of department.)
Avatar of: Mitchell Wachtel

Mitchell Wachtel

Posts: 30

December 18, 2009

It is easier to fake material in immunology is clear; even easier is faking it in studies of human beings, whence concerns, perhaps excessive sometimes, about corporate sponsorship and the like. With structural biology, strict policing is performed by those using the data. If the work is indeed proven fraudulent (not being a jounalist, when my name is mentioned, I have to say things like "if OJ Simpson did indeed murder Nicole Brown Simpson"), what may have happened is a waste of months of work by the Dutch investigators, who then had to spend a month or so figuring out what was going on. Scientists who have wasted months on such shenanigans tend to be very, very angry persons. Fraud in this stuff will definitely get caught within a year or two. For this reason, it is probably not necessary to ramp up the review process a great deal. \n\nClever frauds and criminals always fool those who surround them; think about how long Dr. Swango got away with all those murders. Structural biology is esoteric; highly unusual would be any but another structural biologist's detecting fraud by a structural biologist. There is likely not much UAB could have done.
Avatar of: anonymous poster

anonymous poster

Posts: 2

December 18, 2009

Dr. Thomas. Dr. Contreras. Dr. Murthy. And that's just 2009. The Office of Research Integrity needs to take a good hard look at this institution, both these cases and other cases not forwarded to ORI but which may very well should have been.
Avatar of: Paul Young

Paul Young

Posts: 1

December 19, 2009

It is extraordinary how much impact just a few acts of fraudulent reporting can have on other legitimate investigators. My involvement in this story involves the early fraudulent structures of the dengue protease. Dengue affects some 100 million people globally each year and has been a target for both active vaccine and antiviral programs for many years. The protease is essential for virus growth and so much effort has been put towards the development of inhibitors of this enzyme. The Murthy crystal structure played a very important role in directing much of that early research effort, until legitimate structures for the dengue and West Nile virus proteases were published in 2006 (Erbel et al, Nat.Struct.Mol.Biol. 13, 372, 2006). \n\nOur part in this story started when we published in early 1999 a paper describing a homology model of the dengue protease based on the recent (at the time) report of the HCV protease structure. These viruses are in the same family, Flaviviridae and it seemed like a logical model to generate as the basis for the design of further biochemical studies (Brinkworth et al, JGV, 80, 1167 1999). We were driven to this by our failure in our early attempts to crystallize a recombinant form of the protease. So when the Murthy paper came out later that same year we saw nothing to question as it looked a lot like our model. As it turns out, the UAB investigations have revealed that the Murthy structure could be super-imposed over the HCV crystal structure 1JXP, so it wasn't really surprising!\n\nWe spent a lot of the subsequent year (2000) trying to repeat the Murthy crystallization, even to the point of contacting the researchers directly to get their detailed protocols - which they provided. We followed these to the letter but no matter what we did we couldn't get the protein to crystallize as they had supposedly done. Nevertheless, we took their structure at face value (as no doubt did many others in our field) and designed mutagenesis studies that allowed us to probe structure-activity relationships. We went on to design a novel recombinant form of the protein which was the basis for the subsequent successful crystallizations.\n\nWhen you take the combined research effort and tax-payer dollars (in our case Australian tax-payers!) that went into follow-up work based on this particular component of the fraud perpetrated by Murthy and then add all the stories of the false paths followed by researchers following the other structures you may just start to get an impression of how big a story this is!\n\nWe need to take integrity in research much more seriously. In the past, this essential component of our conduct as scientists has often been embedded in our training as new researchers in a laboratory setting. Perhaps it is time to move discussion and training in these issues into our under-graduate programs and give them the emphasis they need.

December 19, 2009

\n\n** We need to take integrity in research much more seriously. In the past, this essential component of our conduct as scientists has often been embedded in our training as new researchers in a laboratory setting. Perhaps it is time to move discussion and training in these issues into our under-graduate programs and give them the emphasis they need. **\n\nI agree.\n\nIntegrity is intrinsic to the scientific process. Incorrect predictions, misinterpretations, unsubstantiated assumptions have occurred and will continue to happen in science as a result of natural human error. Except that the latter is part of the process and there are internal mechanisms for self-correction and re-direction.\n\nThe emphasis in integrity, as an additional extrinsic element of science, comes to be vital when the scientific process grows misrepresented in its principles, applications and means for conducting it. This misrepresentation can be subconsciously or intentionally effected when the original objectives are subverted as to fit particular interests. Nobody, no organization is immune to fall into the trap. Introspection and discussion of true (and by contrast) distorted processes ought to be programmatic, starting at an early age in the educational system.\n
Avatar of: anonymous poster

anonymous poster

Posts: 1

December 19, 2009

One wonders how this work got past referees since it is claimed that the structure violated elementary protein biophysical principles. Could it be that the referees' judgements were biased by the reputation of the researcher and the institution (who it appears had significant success in acquiring NIH grants)? \nI think the review process should be double blind; that is, the reviewers should not know the name of the PI or the institution the work comes from (PIs already are not privy to the identity of the reviewers). The quality of the work should be judged soley on its merit.\nFinally, it's not clear form the article how the critics are distinguishing between bad scientific methods that just lead to wrong results and out right fraud (faking data). Raw data is never published and many research groups in university settings do not keep detailed and rigorous research notebooks as required in proprietary research settings.
Avatar of: anonymous poster

anonymous poster

Posts: 77

December 19, 2009

Research fraud is treated as the whitest of white-collar crimes and until the perpetrators are punished, as the criminals they are, the rewards will continue to outweigh the risks of detection (obviously, quite low) and punishment by a sizable margin. As almost all these cases involve theft of national resources, under conditions of fraud - e. g., citing previous fraudulent data in support of continuing or new proposals, in the form of national grants, it does not seem a stretch to involve the FBI in tracking down those and itemizing the damage done. \n\nAdditionally, when calculating damages, it must be remembered that, in addition to the intangible damages, such as the reputations of institutions and others in the profession, there are also the "opportunity damages" to those who were denied grants during the same funding cycles in which the fraud was used to deprive other worthy proposals of funding. \n\nTo date, institutions and the law have treated fraudulent scientists apart from fraudulent medical practitioners - the worst of whom are, occasionally, uncovered and prosecuted. The attitude seems to have been that shame and loss of credibility "are punishments enough" and, from the institutional point-of-view, the sooner the problem is disposed of and out of the headlines, the better. \n\nThat approach, quite obviously, is not working and, more realistically and uncomfortably, places the institution in the position of aiding and abetting - with their obvious motive for complicity being continuation of the cash flow to overhead budgets from all grants received. Maybe a good first start, in addition to prosecuting the perpetrator, would be mandatory return of all grant monies, including those that were taken for overhead, by the institutions. That would get their attention in this matter. After all, where else in our society are stolen goods or money allowed to be retained - by innocent parties or not - and not recaptured for the benefit of the original owners - the grant giving agency and, ultimately, the tax payers who generously support honest scientific research.
Avatar of: anonymous poster

anonymous poster

Posts: 1

December 22, 2009

1.Crystallography and protein structure determination are quite specialized fields and it was on the message boards of these fields that the discrepancies in the Murthy structures were posted quite awhile ago. These message boards are not frequented by university administrators, who rely on the peer-review of publications and grant study sections to monitor the quality of research being conducted by their faculty. As soon as the UAB administration was contacted an investigation was initiated.\n2.Integrity in research is essential, however honesty is a character trait that is developed at an early age, so whether or not you get "integrity training" in college or grad school, if you are a dishonest person to begin with it won't make a difference. Honesty and integrity should begin in the home and supported by schools throughout the formative years. The USA is not the only country that needs to emphasize research integrity. It should also be a mandatory part of scientific training in the rest of the world's research community, as this is not behavior that is unique to US scientists. In a world of shrinking research funds and available positions, the competition has made some lose sight of the true beauty of science: the journey one embarks on in the search for answers to complex questions.
Avatar of: anonymous poster

anonymous poster

Posts: 5

December 23, 2009

Notice the common thread for all the tainted structures?\n\nThe UAB page lists the analysis, this being true in every case:\n\n"No raw crystallographic data, data reduction output, or any other experimental records that would support the correctness of the structure of **** or demonstrate that these was an experimentally determined structure, were available for examination."\n\nNow that the PDB requires the deposition of raw (structure factor) data hopefully such situations may be avoided. Additionally, making the raw data available to the referees for peer-review should be made mandatory. \n\n2c
Avatar of: anonymous poster

anonymous poster

Posts: 1

January 31, 2010

Note that structure factors, now required by PDB to be submitted together with the model (the PDB file), does not constitute "raw data". In the case of the Murthy structures, some of them did have structure factors associated with them, but these had been fabricated from the model, which is how the fraud was discovered in the first place. Only raw diffraction images constitute raw data. These are far more difficult to fabricate, but currently there is no requirement for submission of this data, in part due to the large size of the data sets. Also, more frequently, this data is collected remotely at synchrotron sources, but the synchrotrons also do not keep the raw data for very long. This probably ought to change now that data storage is pretty cheap.

Follow The Scientist

icon-facebook icon-linkedin icon-twitter icon-vimeo icon-youtube
Advertisement
EMD Millipore
EMD Millipore

Stay Connected with The Scientist

  • icon-facebook The Scientist Magazine
  • icon-facebook The Scientist Careers
  • icon-facebook Neuroscience Research Techniques
  • icon-facebook Genetic Research Techniques
  • icon-facebook Cell Culture Techniques
  • icon-facebook Microbiology and Immunology
  • icon-facebook Cancer Research and Technology
  • icon-facebook Stem Cell and Regenerative Science
Advertisement
Thermo Scientific
Thermo Scientific
Advertisement
NeuroScientistNews
NeuroScientistNews
Life Technologies