All genes
Early release. This is the first public version — tell us what you find.
Free browser extension

See which antibodies in a paper have been tested

Marks every antibody cited on a page with what independent, knockout-controlled testing found — for the application it was actually used in. Hover any mention to see the data behind it.

The antibodies we have characterised have been used in more than 29,000 published papers, according to CiteAb — the extension tells you which ones you are looking at.

Works in Chrome, Edge, Brave and Firefox. Free, and always will be.

What it looks like on a paper

Real papers, marked by the extension. Each mention is coloured for the application that paper used it in; hovering one opens the card behind it, with the knockout-controlled image the result rests on.

A mark is a result, not advice: each one reports how that antibody performed when it was tested against a genetic control under the community consensus protocols, in the applications those protocols cover.

A hover card over a methods section, showing GTX637119 recommended for western blot, above a knockout-controlled blot.
Recommended — and the blot behind it. GTX637119 on a Europe PMC methods section: recommended for western blot under the consensus protocols, with the wild-type / SYT1-knockout blot the result rests on.
The same card with the immunoprecipitation tab selected, showing starting material, unbound and IP lanes.
One result per application. The same antibody, immunoprecipitation selected. Not recommended for IP in the conditions tested — and the SM / UB / IP lanes behind that call are shown, not just asserted.
The card on its immunofluorescence tab, a tooltip naming the verdict, and a note that results are fixation and permeabilisation dependent.
Hover a tile, get the verdict. Immunofluorescence: not recommended. The wild-type / knockout image is still shown, so you can judge the call rather than take it on trust.
The card on its flow cytometry tab, showing knockout and wild-type populations overlaid.
Flow cytometry, same standard. Not recommended for FC in the conditions tested, with the histogram against the knockout control and the note that performance is protocol and sample dependent.
A card recommending an antibody for three applications, with the fourth tab stating no immunofluorescence figure has been published.
Recommended for three applications. MAB4364 is recommended for WB, IP and FC. IF has not been tested — the extension says so plainly instead of leaving a gap you might read as a pass.
A card over the “Synaptotagmin-1 antibodies” phrase, offering characterised SYT1 antibodies instead.
No catalogue number? Still useful.. When a paper names only the target, the extension points to the antibodies that do have knockout-controlled data — and says plainly that this one has not been tested.
The extension popup showing counts by verdict and a list of antibodies not recommended for the applications used.
The whole paper at a glance. Every antibody mention on the page, counted: recommended, not recommended, split by application, and how many have tested alternatives. Click a count to jump through those marks.

What the colours mean

Results are given per application. An antibody that works on a Western blot may fail in immunofluorescence, so the extension reads the application from the surrounding text and colours the mention for that application. Of the antibodies we have published against TDP-43, two thirds pass in one application and fail in another.

ColourMeaning
Recommended Tested with knockout controls under the consensus protocols, and recommended for the application used here in the conditions tested.
Not recommended Tested with knockout controls under the consensus protocols, and did not meet the bar for the application used here in the conditions tested. Where the page doesn't say which application, this means it did not meet the bar in any application tested. It is a result under one protocol in one cell line, not a judgement on the reagent everywhere.
Split Recommended in some of the applications tested and not in others. Either the paper used it for several applications with a different result in each, or it never said which application it used, so all four assessed are shown. The hover card says which, and lists the results.
Alternatives exist This antibody has not been tested, but its target has. Validated alternatives are available.
Application untested In the dataset, but not tested in the application this page used it for. Antibodies we hold no data on are left unmarked.

Results are based on consensus protocols. Antibody performance is protocol and sample dependent, and these results do not validate or invalidate experiments in other assay systems or sample types. Read the protocols.

Immunofluorescence results are fixation and permeabilisation dependent.

It does not watch you read

The dataset is downloaded and stored in your browser, and all matching happens there. No page text, URL or reading history is sent to us or to anyone else. An unpublished or under-review manuscript stays private.

The extension contacts this site for exactly two things: a once-a-day refresh of the validation data, and loading a validation image when you open a card. Neither carries any information about what you are reading.

If your institution blocks extensions

Many universities and institutes manage Chrome, Edge and Firefox centrally and allow only approved extensions. If Add to Chrome does nothing, or Firefox refuses the install, that is usually what has happened — and it is a decision your IT team can reverse. They generally need three things: what it does, what it can reach, and its identifier.

A note you can send them

I'd like to install a browser extension called Only Good Antibodies — validation at a glance. It marks antibodies cited in a paper with what independent, knockout-controlled testing found, which helps me avoid buying reagents that have been shown not to work.

It stores its dataset locally and does all matching in the browser. No page text, URL or reading history is sent anywhere — the manifest declares no data collection. It contacts one domain, onlygoodantibodies.co.uk, about twice a day to refresh that dataset and to load an image when I open a card.

It asks for three permissions — storage, alarms and scripting — and by default runs only on a published list of 196 journal and preprint domains. It has no access to tabs, cookies, browsing history or downloads. Anything wider is optional and off unless I turn it on.

The details you'll need to allow it are at https://onlygoodantibodies.co.uk/extension/.

What IT will ask for

Chrome / Edge ID pkmfholpagoaiigfpbkalnpipkiiidlp
Firefox add-on ID extension@onlygoodantibodies.co.uk
Firefox install URL https://onlygoodantibodies.co.uk/extension/firefox.xpi

Mozilla-signed. Updates come from https://onlygoodantibodies.co.uk/extension/updates.json.

Permissions storage, alarms, scripting — and nothing else. No tabs, cookies, history or downloads.
Sites it runs on A named list of 196 journal and preprint domains, in the manifest. Supplier catalogue pages and “any site” are optional permissions, off until the reader switches them on.
Data collected None. The manifest declares no data collection, and matching happens entirely in the browser.
Network access One domain: https://onlygoodantibodies.co.uk/*. A daily dataset refresh and an image when a card is opened.

The policy settings

For an administrator who already manages browsers. Allowing lets people install it if they want to; force-installing deploys it to everyone. Either is fine — nothing about the extension needs it to be mandatory.

Chrome and Edge — add the ID to ExtensionInstallAllowlist, or to ExtensionInstallForcelist to deploy it:

ExtensionInstallAllowlist = [ "pkmfholpagoaiigfpbkalnpipkiiidlp" ]

Firefox — an ExtensionSettings entry. The extension is distributed by us rather than through addons.mozilla.org, so the install URL is ours:

"ExtensionSettings": {
  "extension@onlygoodantibodies.co.uk": {
    "installation_mode": "allowed",
    "install_url": "https://onlygoodantibodies.co.uk/extension/firefox.xpi"
  }
}

Use force_installed in place of allowed to deploy it to everyone. Firefox needs install_url either way, because the add-on is not listed on addons.mozilla.org.

If the answer is no

Nothing here is only in the extension. Every verdict it shows is on the gene pages of this site, in the public API, and through the MCP connector, which needs no browser extension at all. The extension saves you looking things up; it is not the only way to get the answer.

Where it runs

On journal and preprint sites — Nature, Science, Cell Press, ScienceDirect, Wiley, Springer, PLOS, eLife, Frontiers, MDPI, OUP, bioRxiv, medRxiv, PubMed and PMC among others. Supplier catalogue pages can be switched on in settings, so you see the same results while choosing what to buy.

It reads the HTML full text of a page. A paper opened as a PDF is not annotated.

What it can and cannot tell you

It is an early release — tell us what you find

This is the first public version. The most useful thing you can do is open papers you know well and check the marks against what you already believe about those antibodies. Worth a message:

Say which page you were on and what you expected — get in touch. A store review helps other people find the extension; the specifics reach us faster here.

Building on the same data

The extension reads a public snapshot of the dataset, refreshed daily:

/extension/index.json

The same data is available to AI assistants through our MCP connector, so an assistant reading a manuscript can flag the same antibodies. If you are building something with it, tell us — we would like to help.

Spotted an antibody we haven't tested?

Nominate a gene