One postdoc found 17,000 lab antibodies sold with doctored proof images

The Western blot was running on a Tuesday afternoon, the kind of routine experiment that every biology lab in the world runs without thinking. A researcher pipettes a sample, runs it through a gel, transfers it to a membrane, and adds an antibody, a Y-shaped protein that latches onto one specific target in the sample and lights it up. The "proof image" the lab trusts to show whether the antibody worked is a photograph of that lit-up band. Most of those antibodies, the ones on every bench scientist's order list, cost around $300 and arrive with a glossy vendor photo showing they did their job.

The image the vendor ships is exactly what Reese Richardson has spent the past three months learning to distrust., and has since expanded the search. What began as a single-vendor complaint about Thermo Fisher's antibody catalog has, according to an investigation published Tuesday by Ars Technica, grown into a sweeping finding: more than 17,000 commercial antibody products across at least 15 companies are now listed in Richardson's public audit as carrying problematic marketing images.

Reporting from Nature puts the image count higher still, with more than 18,000 questionable images across the catalogs of those 15 firms. Thermo Fisher's catalog alone contains more than 100 suspicious images flagged by science sleuths, per C&EN's earlier coverage.

Some are cosmetic: a vendor cleaned up background noise, bumped contrast, or cropped a lane to make a blot look tidier than the raw capture. Others are not cosmetic at all. Richardson's audit, mirrored on his public blog, documents copy-and-paste jobs in which bands from one experiment have been lifted into another, and outright fabrication in which validation data appears to have been generated rather than recorded. In an academic paper, that second category would trigger a retraction. In a vendor catalog, it has been a marketing practice.

This is not a story about bench scientists being fooled. The protein-detection experiments a working biologist runs with a vendor antibody (Western blot, immunofluorescence, the tissue-staining protocols that map where a protein lives in a cell) depend on a chain of trust that runs from the antibody maker to the bench. If the proof image is fabricated, the scientist had no practical way to know before ordering. The audit is the first time that chain has been pulled into the open.

The Scientist, a life-sciences trade publication, reports that independent researchers have begun flagging the same patterns, and that vendors are starting to respond. None of the affected companies have publicly conceded that the audit's image-pattern heuristics are catching real fabrications rather than marketing flourishes, but several have begun pulling listings while they review.

The constructive exit here is narrow but real. Richardson's audit is public. A lab can search it before ordering. A journal reviewer can ask whether the antibody in a submitted figure has a clean audit entry. The mechanism is small (one researcher, an image-pattern search, a public spreadsheet), but it has converted a hidden, normalized practice into a record that anyone can check. The next question is whether the vendors treat it as a starting point or as a public-relations problem.