More than 18,000 questionable images found in antibody catalogues of 15 companies
核心洞察
A metascientist identified more than 18,000 digitally altered validation images across antibody catalogues from 15 commercial vendors, raising concerns about supplier practices.
Four companies account for the bulk of flagged images: G-Biosciences (搜索) (6,388), Thermo Fisher (5,580), LSBio (搜索) (2,695), and Origene (搜索) (2,236).
The findings follow a 2023 study showing over 50% of 614 commercial antibodies failed to work as advertised, with ineffective antibodies costing more than US$1 billion annually.
Thousands of antibodies sold by multiple research vendors have images associated with their catalogue entries that appear to have been digitally altered, according to a scientific-integrity specialist. The finding, reported on 25 August by Reese Richardson, a metascientist at Northwestern University in Evanston, Illinois, identified more than 18,000 questionable images associated with antibodies sold by 15 companies, and released a database of them.
The latest search follows an earlier inquiry in May, when science sleuths uncovered more than 100 suspicious images in the antibody catalogue of supply company Thermo Fisher Scientific (搜索). Although image alteration does not mean an underlying product is defective, the trove of flagged images raises questions about the practices of commercial antibody suppliers as the scientific community grapples with concerns about the reliability of these workhorse diagnostic molecules.
Many of the "validation images," which are included in catalogue entries to confirm antibodies' quality, show signs of extensive editing, according to Richardson.
The scale of the problem
Antibodies are Y-shaped biomolecules that can bind to specific proteins, making them crucial diagnostic tools in the laboratory. However, their reliability has been a persistent problem; a 2023 study examining 614 commercial antibodies showed that more than 50% of them did not work as advertised. The annual cost of ineffective antibodies is estimated to be more than US$1 billion.
Most of the vendors in Richardson's database are each associated with a few hundred potentially problematic images. But four are tied to thousands of flagged validation images: G-Biosciences (搜索) in St. Louis, Missouri, is linked with 6,388 images; Thermo Fisher in Waltham, Massachusetts, with 5,580; LSBio (搜索) in Newark, California, with 2,695; and Origene (搜索), in Rockville, Maryland, with 2,236.
Researchers who work on improving antibody reliability are frustrated by the news. "It's disappointing, but it doesn't surprise me that much because I've been burned so many times buying commercial antibodies that don't work," says Simona Colantonio, director of the Antibody Characterization Laboratory at the Frederick National Laboratory for Cancer Research in Maryland.
Vendor responses
A spokesperson for Thermo Fisher said in a statement that an ongoing review of the firm's products had so far found "no evidence of impact to antibody performance or the underlying scientific data" and that the company remains "confident in the quality of the products within our antibody catalog."
A spokesperson for LSBio (搜索) said in a statement that the entries "amount to less than 0.04% of the total offering on our platform," but that the firm takes the image concerns seriously and will perform an investigation with its partners. G-Biosciences (搜索) and Origene (搜索) did not respond to a request for comment.
Cataloguing issues and methodology
Researchers looking to purchase antibodies, much like other online shoppers, sift through lengthy catalogues before making their choice; some vendors offer hundreds of thousands of antibodies. Companies typically include image data such as western blots, in which a specific protein in a mixture is "labelled" by a given antibody. Antibodies that perform well leave a distinct band in a blot, and those that perform poorly, by failing to bind to their specific target, can have multiple fuzzy bands.
These images are "really one of the few things you have to go on" when determining whether an antibody is worth purchasing, says Sholto David, a molecular biologist in Oxford, UK, who specializes in research-integrity issues.
When David and Richardson identified more than 100 suspect images in the catalogue of Thermo Fisher in May, the firm responded publicly by saying that it "fully stands by the data and underlying science" but noted that image data might have been "edited, optimized or otherwise adjusted" for presentation.
To investigate further, Richardson scanned the catalogue of Thermo Fisher more thoroughly and looked at those of other companies. He wrote a program to automatically raise and lower the black and white levels in more than 270,000 images, to increase the contrast and reveal any alterations. Then, he checked each one manually.
The 18,000-plus images Richardson has now flagged have not all been double-checked by independent parties, but the handful of fellow sleuths and researchers whom Nature asked to examine a dossier of Richardson's findings say that his method seems sound. "Some of them are really outrageous because you don't even need to increase the contrast to see that there is some manipulation," Colantonio says. "The sad part is some of the antibodies are actually probably very good" and these alterations could place doubt on their performance, she adds.
A system-wide concern
The sheer number of questionable images flagged by Richardson has alarmed researchers. "What we're actually looking at is a system-wide problem that requires a system-wide solution," says pulmonologist Harvinder Virk, co-founder of Only Good Antibodies in Leicester, UK, an open-science effort to characterize and validate commercially sold antibodies.