Rune Busk Damgaard

More evidence that LUBAC has functions beyond immune signalling. Our recent study shows that LUBAC and OTULIN regulate metabolic signalling: specifically they govern AMPK signalling to control metabolic adaption, autophagy, and cell death during energetic stress. This suggests that the metabolic symptoms observed in patients with LUBAC deficiency and ORAS, including glycogen storage disease, lipodystrophy, and liver disease, may be caused directly my metabolic dysregulation rather than being secondary to the systemic inflammation. nature.com/articles/s4141…

A modest proposal: If you submit a manuscript with a made-up citation—not a bungled or misspelled reference—you receive a 5-year ban on having your work reviewed at that journal.

One of my NIH grant reviews came back last year with comments clearly LLM-generated. The PMIDs cited against the proposal were hallucinated. The papers did not exist. The score still counted toward the funding decision.

"Mens Alternativets forperson i København, Birgitte Kehler Holst, var på ferie i Chile i december og januar. ." Og det var Chile. Svømmere hun mon derned, siden hun tillader sig at være så hellig på de ældres bekostning? ekstrabladet.dk/nyheder/politi…

ERC is at a turning point, and not in the right direction. Excellent science needs support, not barriers @ERC_Research Read & sign 👇 docs.google.com/forms/d/e/1FAI…

Without Xrays there are no broken bones

Academia feels increasingly like playing Monkey Island, if you miss something as absurd as picking up a rubber chicken with a pulley in the first scene, you only realize much later that you’re stuck and can’t pass to the next stage. I was interviewed today by @ScienceMagazine about the funding crisis following the ERC announcement. I argued that when a grant rejection can hinge on something small, and you then have to wait years to resubmit, the cost becomes so high that instead of suggesting “high-risk high-gain” plans scientists will prepare more cautious, less ambitious proposals. It turns into a perverse game of simplifying your ideas and trying (unsuccessfully) to predict what an overstretched reviewer pool will find easiest to digest. "Abrupt change to European funder’s rules leaves researchers shut out | Science | AAAS" science.org/content/articl…

@WalentekLab @MicrobiomDigest But we may also just have to disagree...

@WalentekLab @MicrobiomDigest I think it depends on our definitions maybe. You seem to argue based on how easy it is to produce fake data de novo? My argument is that you can as easily make fraudulent data by manipulating microscopy images, either by photoshopping, biased selection, or false labelling.

I knew this was coming. We are so f*cked - fake and real now look the same. We desperately need tools to show journals and researchers that data is real, not AI-generated.

@WalentekLab @MicrobiomDigest Dont agree. I respect your opinion, but I think it is prejudiced and circular. Sleuths identify many fake/photoshopped tissue sections. And just look at how far AI has come in 2-3 years. Imagine in 10. You have just decided the standard is different for your favourite data.

@rbdamgaard @MicrobiomDigest Nah. Not for a while. While WBs were already faked before photoshop. 🤷‍♂️

@WalentekLab @MicrobiomDigest For now maybe. Maybe in a few months it’s not when the next image generator is released. Then what?

@rbdamgaard @MicrobiomDigest What? Now you're teasing me. Complex tissue images are Much harder to fake than bands.

@WalentekLab @MicrobiomDigest How would you show a morphological phenotype? A photograph or micrograph? Can’t that be faked by AI the same way a western blot can?

@rbdamgaard @MicrobiomDigest A morphological phenotype for example.

@WalentekLab @MicrobiomDigest But what kind of data?

@rbdamgaard @MicrobiomDigest Supporting orthogonal experiments to validate a core concept of study. As stated in the initial reply of mine.

@WalentekLab @MicrobiomDigest For what?

@rbdamgaard @MicrobiomDigest As supporting info.

@WalentekLab @MicrobiomDigest You write “so I stopped believing in them as primary source for validating a concept”. I’m not sure ai understand how that is misaligned witt what I wrote? In what context do you believe in them then?

@rbdamgaard @MicrobiomDigest Read carefully. Not what I said.

@aykutuz But I agree we should be more accepting of negative results.

Thanks for the reply. The real reason: Scientific work is messy. You try one thing and mess it up, then try another. Sometimes you didn’t set up the environment correctly and the measurement ends up being garbage. The scientist then feels justified in eliminating these data. That’s somewhat understandable. Sometimes the measurements give results that really don’t support the main thesis, while the majority do. One may think, “Perhaps this is because of the environment setup that I did incorrectly. I shouldn’t let these points sabotage our work - especially since other people are doing the same.” This is problematic. Spherical chickens are fine as assumptions, but this shouldn’t extend into scientific results. How can we understand/infer there's such an inclination in “scientific” publications?: We don’t see nearly enough “negative” result papers: “We tried this and it didn’t turn out as we expected.” That is fine, it is science. It is as valid as work that supports a thesis. These papers should have outnumbered the positive ones, but they do not. So we can infer a motive of “sphericalization” of experimental results. A little into the modus operandi: What happens to these unpublished works? The model is usually updated only until a new set of experiments arouse that support a version. That introduces a subtle, systemic selection bias into scientific work. In a scientific paper, I demand: 1. an abstract that has a neutral statement vis-à-vis the thesis, 2. which should be submitted “before” the experiments, and 3. experiment data which is messy. 4. The body should mainly explain the model/thesis, not attempt to prove it. As a result, I expect to see many “negative” result scientific papers. If you agree with any of the above, it is your responsibility to steer the current way “scientific” publication works toward how science should ideally work.

Because most scientists did not become scientists to fake data and be fraudsters. Most scientists are honest and decent.

@aykutuz Your demands may work in computer science, but that’s not how it works in biology.

I'll bet against this. The foundations of science have never been about trust (nullius in verba, etc), and convincing media is not the ground truth of our trust network. That our scientific institutions use certain trust mechanisms for efficiency reasons does not mean the field melts when those specific mechanisms no longer work, nor does it mean we have to learn to survive in a trustless dystopia. We just develop new trust mechanisms. This sort of alarmism is analogous to all the hand-wringing over deep fakes, which were predictably a nothing burger. blog.jessriedel.com/2018/04/23/meh…

@aykutuz And for some data types, it’s a requirement and people share the raw data.

@aykutuz There are at least 3 reasons: 1) it’s not required by most journals, 2) it’s an enormous effort and extra cost, 3) it provides opportunity for people working in bad faith to try and undermine the work. And there are probably more reasons than that.

@aykutuz In principle, yes.

@rbdamgaard Let’s assume this is true. Then “most” should be happy if raw data availability becomes the norm; to make “minority “ fraudsters’ job difficult; right?