A postdoc spends four months optimizing a Western blot for a tumor-suppressor target. The band is crisp, the figure looks publishable — then a knockout control lights up just as brightly. The antibody was never specific. The reagent that quietly decided the result was the one nobody had validated.
Scenes like this are neither rare nor cheap, and they are largely avoidable. The questions that separate a trustworthy antibody from a risky one are few, concrete, and answerable from the datasheet before you spend a cent.
The reagent most likely to cost you months
Antibodies are among the most widely used and least scrutinized reagents in cancer research, and when experiments fail to reproduce they are a leading reason why. In a Nature survey of 1,576 researchers, more than 70% had failed to reproduce another scientist's experiments and more than half had failed to reproduce their own.1 An economic analysis estimated that irreproducible preclinical research costs roughly US$28 billion a year in the United States alone, with biological reagents and reference materials among the largest contributors.2 For antibodies specifically, an estimated US$800 million is wasted worldwide every year on reagents that do not perform as described.3
The encouraging part is that antibody failures are systematic, not mysterious. They trace back to a short list of causes you can screen for in advance.
Why two antibodies to the same target are not the same
Reach for two catalog antibodies against the same protein and you may get two different results — both "correct." They can recognize different epitopes, tolerate fixation differently, and perform in one application while failing in another. A denatured epitope on a Western blot membrane is not the native, folded epitope an antibody must find in fixed tissue (IHC) or intact cells (IF/flow). Polyclonal reagents add a second variable: because each bleed is a different mixture of antibodies, performance can drift from lot to lot. And a single clean band tells you less than it seems — without a specificity control, you cannot know whether it is your target or a same-size off-target protein.
What “validated” actually means
"Validated" is not a marketing word; it has a working definition. The International Working Group for Antibody Validation (IWGAV) proposed five conceptual pillars for antibody validation — and, critically, validation must be demonstrated for each application separately rather than assumed to transfer.4
- Genetic — signal disappears in a knockout or knockdown (CRISPR/RNAi) control. The strongest evidence of specificity.
- Orthogonal — antibody signal correlates with an antibody-independent measure of the target (e.g. RNA-seq or mass spectrometry across samples).
- Independent antibody — a second, non-overlapping antibody to the same target gives a concordant result.
- Recombinant expression — signal tracks the over-expression of a tagged version of the target.
- Immunocapture mass spectrometry — the antibody pulls down the intended protein, confirmed by MS.
You rarely need all five. One or two well-chosen pillars, matched to the application you will actually run, is the practical bar — and it is already more evidence than most catalog antibodies carry.
Five questions to ask before you buy
1. Is it validated in the application you will run?
An antibody validated for Western blot is not automatically valid for IHC, IF or flow cytometry. Look for application-specific data — ideally an image — for your exact technique. If the datasheet shows only a WB and you need IHC, treat it as unvalidated for IHC.
2. Does it react with your species?
Confirmed reactivity (human, mouse, rat) should be stated and, ideally, demonstrated. Cross-species inference from sequence homology is a hypothesis, not validation.
3. What specificity controls exist?
This is where the pillars earn their keep. A genetic knockout/knockdown control is the gold standard; peptide-block, over-expression and isotype controls add support. For a Western blot, a band at the expected molecular weight with minimal off-target signal is the basic, checkable floor.
4. Is lot-to-lot performance documented?
Polyclonals can drift between lots. A Certificate of Analysis (CoA) and a documented lot history let you trust that this vial behaves like the one in the published figure. A sequence-defined recombinant or monoclonal removes most of this risk.
5. Is there independent evidence of use?
Citations and third-party data are supporting evidence — useful, but secondary to application- and species-matched validation for your own workflow.
Match the antibody to your readout
Decide the application first, then choose the antibody validated for it. A monoclonal offers lot-to-lot consistency and epitope precision; a well-characterized polyclonal can give stronger signal across several applications. Neither is universally "better" — the right choice depends on your readout and whether you need one application or many.
Validated antibodies across cancer mechanisms
It helps to organize antibody selection by the tumor mechanism you are studying rather than by a long catalog. A few examples of application-validated targets, each backed by datasheet validation data:
- Signaling & EMT — YAP1 (Hippo pathway) and SMAD3 (TGF-β).
- DNA-damage response — XRCC4 (non-homologous end joining).
- Tumor metabolism — GLUT1 / SLC2A1 (the Warburg phenotype).
- Tumor suppressor / exosomes — TSG101.
- Microenvironment & ECM — SPARC and Versican.
How to read a BioHippo antibody datasheet
Each listing states the validated applications, confirmed species reactivity, clonality and host, available conjugates, and — where available — a validation image and Certificate of Analysis. Read those fields against the five questions above before adding to cart.
Start with Validated Cancer Antibodies
Our Validated Cancer Antibodies collection gathers antibodies validated in two or more applications (Western blot, IHC, immunofluorescence) to high-value cancer targets. Browse the wider Cancer Antibodies catalog and save 15% with code ONCO15. Validation is per-clone and per-lot — always confirm the applications and images on the specific datasheet.
References
- Baker M. 1,500 scientists lift the lid on reproducibility. Nature 533, 452–454 (2016). nature.com/articles/533452a
- Freedman LP, Cockburn IM, Simcoe TS. The Economics of Reproducibility in Preclinical Research. PLOS Biology 13(6): e1002165 (2015). journals.plos.org
- Bradbury A, Plückthun A. Reproducibility: Standardize antibodies used in research. Nature 518, 27–29 (2015). nature.com/articles/518027a
- Uhlen M, et al. A proposal for validation of antibodies. Nature Methods 13, 823–827 (2016). nature.com/articles/nmeth.3995