More than 300 published papers, including studies appearing in journals such as Nature Medicine, Cancer Cell, eLife, Science Advances and the Journal of Biological Chemistry, may have used antibodies against the wrong protein.
This is not a hypothetical example of poor laboratory practice. It is the result of an investigation by independent researcher Sholto David into a widespread case of mistaken protein identity involving p16-INK4a and p16-ARC. The issue was subsequently reported by Science in June 2026 (Mallapaty, 2026; David, 2026). (Science)
Researchers intending to study p16-INK4a, an important tumour suppressor and widely used marker of cellular senescence, had instead ordered and used antibodies targeting p16-ARC, also known as ARPC5.
Despite sharing the shorthand name “p16”, the two proteins are biologically unrelated.
p16-INK4a is encoded by the CDKN2A gene and helps regulate cell-cycle progression. By contrast, p16-ARC is a subunit of the Arp2/3 complex involved in actin polymerisation and cytoskeletal organisation (UniProt Consortium, 2026a; Abcam, 2026a). (UniProt)
One helps regulate whether a cell continues dividing.
The other contributes to the structure and movement of the cell.
The similarity largely ends with the name.
How citation count became a substitute for evidence
Antibody purchasing often follows a familiar pattern.
A researcher identifies a target, searches supplier catalogues and checks whether an antibody has appeared in previously published studies. A large citation count can create a strong impression of reliability. It suggests that other laboratories have used the product successfully and that the resulting data have survived peer review.
However, citation frequency does not demonstrate target specificity.
In the p16 case, David identified repeated incorrect use of three antibodies:
-
Abcam ab51243
-
Abcam ab151303
-
Santa Cruz Biotechnology sc-166760
These products target p16-ARC or ARPC5, not CDKN2A/p16-INK4a. David reported that approximately 95% of the papers he assessed had used the antibodies incorrectly, while only 17 papers appeared to have used them for their intended ARPC5 target (David, 2026). (For Better Science)
Abcam’s current product information is explicit. The pages for ab51243 and ab151303 state that the antibodies detect p16-ARC and do not detect CDKN2A/p16-INK4a. Abcam also acknowledges that some publications have incorrectly described these products as detecting p16-INK4a (Abcam, 2026a; Abcam, 2026b). (Abcam)
The products were therefore not necessarily mislabelled by the manufacturer. The central failure occurred when researchers selected, cited and used antibodies without confirming that the catalogue number corresponded to the intended biological target.
Why the wrong antibody could still produce convincing data
The mistake was able to spread because an incorrect antibody does not always produce an obviously incorrect-looking result.
ARPC5 is associated with the actin cytoskeleton and is expressed across many cell types. An antibody detecting this protein may therefore generate clear and repeatable bands in a western blot.
Those bands can look technically convincing.
They can appear near an expected molecular weight.
They can vary between samples.
They can even correlate with broader changes in cell structure, protein loading or experimental conditions.
But a clear band is not evidence that the correct protein has been detected.
David suggested that p16-ARC expression may, in some experimental contexts, behave more like an actin-associated loading signal than a genuine measurement of p16-INK4a. This creates an especially dangerous situation: the wrong antibody can generate data that appear internally consistent and can therefore reinforce the researcher’s expectations (David, 2026). (For Better Science)
Once those findings are published, later researchers may interpret the publication itself as evidence that the antibody is suitable.
The error then becomes self-reinforcing:
-
A paper uses the wrong antibody.
-
The paper reports apparently plausible results.
-
Other researchers identify the catalogue number in the methods.
-
They purchase the same product.
-
The growing citation count increases confidence in the antibody.
-
The original mistake spreads through the literature.
Citation volume becomes a proxy for validation, even though every new citation may be reproducing the same unresolved error.
The controls that could have identified the problem
The p16 mix-up could have been detected using routine target-specific controls.
One approach would have been to test the antibody in a cell line carrying a homozygous deletion of CDKN2A. Such a model cannot produce p16-INK4a. If the antibody continued to generate the same band in the knockout or null sample, this would demonstrate that the signal was not dependent on the intended target.
A second approach would have been to reduce CDKN2A expression using siRNA or another genetic perturbation method. A target-specific antibody signal should decrease in line with confirmed target depletion.
These experiments address a fundamental question:
Does the detected signal disappear when the target is absent?
A band at the expected molecular weight is not enough.
A supplier image is not enough.
A highly cited paper is not enough.
Even a reproducible signal is not enough if the identity of that signal has not been established.
The current ab51243 product page reports that the antibody has now been validated using knockout samples for its correct target, ARPC5. This supports its specificity for p16-ARC, but it also reinforces why the antibody cannot be treated as a CDKN2A/p16-INK4a reagent (Abcam, 2026a). (Abcam)
This was not simply a supplier failure
It would be easy to describe the p16 case as another example of a manufacturer selling a poor antibody.
That would miss the central lesson.
The antibodies appear to have been designed for and marketed against p16-ARC. The problem arose because researchers confused p16-ARC with p16-INK4a, failed to confirm the target identity associated with the catalogue number and then cited earlier incorrect use as evidence of suitability.
Responsibility therefore extends across the research system.
Researchers must confirm product identity and perform appropriate controls.
Suppliers must make target names, gene identifiers, immunogens and known limitations unambiguous.
Reviewers and journals must examine whether key reagents are appropriate for the claims being made.
Procurement systems must stop treating catalogue familiarity and citation volume as substitutes for evidence.
Databases and reagent registries must make known concerns visible before another laboratory purchases the same product.
The Research Resource Identification initiative has begun attaching warnings to affected antibody records, demonstrating how persistent identifiers and centralised reagent records can help expose recurring problems across the literature (RRID, 2026). (rrid.site)
What the case reveals about scientific reproducibility
The p16 incident demonstrates that reproducibility cannot be addressed solely by asking researchers to write more detailed methods sections.
A study may report its antibody supplier, catalogue number, dilution, incubation time and detection method perfectly.
It may still be biologically wrong.
Transparent reporting makes an experiment easier to repeat. It does not guarantee that the underlying reagent was appropriate.
The distinction matters.
Repeatability asks whether another researcher can follow the same procedure and obtain a similar signal.
Validity asks whether that signal represents the biological target being claimed.
The p16 papers may, in some cases, have produced repeatable ARPC5 signals. That does not make those signals valid measurements of p16-INK4a.
This is why reagent identification and reagent validation must be treated as separate requirements.
Identification confirms what product was used.
Validation establishes what the product actually detects under the relevant experimental conditions.
What ABMIUM stands for
ABMIUM was built around a simple principle:
A reagent should not be trusted because it is popular. It should be trusted because the evidence supports its intended use.
Our verification-led model examines more than citation count or supplier reputation. We assess factors including product provenance, manufacturer transparency, documentation, quality-control ownership and the strength of the available application-specific evidence.
Through ABMIUM Verified™, we identify products from original manufacturers whose traceability, documentation and commercial claims have been assessed.
Through ABMIUM Validated™, selected products undergo defined, application-specific evaluation supported by independent scientific partners.
These are deliberately different standards.
Supplier verification does not automatically prove that every antibody will work in every sample type or application. Equally, a successful western blot does not establish suitability for immunohistochemistry, immunoprecipitation or immunofluorescence.
Each claim must be linked to the evidence that supports it.
Where appropriate, stronger antibody validation should include:
-
genetically defined positive and negative controls
-
knockout or knockdown samples
-
orthogonal confirmation using a non-antibody method
-
relevant sample types and preparation conditions
-
assessment of unexpected bands or staining
-
lot and clone traceability
-
application-specific performance data
-
transparent reporting of negative and inconclusive results
Validation should not be treated as a badge added after the product has already entered widespread use. It should be the process through which a specificity claim earns scientific credibility.
The broader warning for the field
The p16 case is unlikely to be unique.
David has also highlighted confusion involving the abbreviation PDK1, which can refer to different proteins depending on the field and naming convention. Such examples suggest that ambiguous protein names, incomplete reagent identifiers and inherited citation practices can create systematic errors across entirely separate areas of biology (David, 2026). (For Better Science)
The wider problem is not that researchers lack intelligence or care.
It is that the research system often rewards speed, recognisable suppliers and publication precedent more strongly than careful reagent qualification.
A reagent already used in 100 papers may feel safer than one supported by stronger but less visible validation data.
The p16 case demonstrates why the opposite may be true.
A large citation count can reflect widespread confidence.
It can also reflect widespread repetition of the same mistake.
What this means when sourcing through ABMIUM
ABMIUM does not treat publication count as proof of target specificity.
We use citation evidence as one part of a wider assessment, not as the final answer. Products and suppliers are considered in relation to their provenance, documentation, validation evidence, application claims and control of manufacturing and quality processes.
Where independent validation has been completed, the claim must remain tied to the precise conditions tested.
That means identifying:
-
the target
-
the application
-
the species
-
the sample type
-
the controls
-
the observed result
-
the limitations of the evidence
No antibody is universally validated simply because it worked once.
No reagent becomes reliable merely because a prestigious journal published a study using it.
And no catalogue number should be copied into a new protocol without confirming what the product actually detects.
The p16 antibody mix-up is an extreme example, but the behaviour that allowed it to spread is routine: selecting reagents through citation inheritance, assuming that a familiar supplier guarantees suitability and accepting plausible-looking signals without establishing target dependence.
Science advances on the quality of its reagents as much as it advances on the quality of its ideas.
ABMIUM exists to help make that quality verifiable, not assumed.
References
Abcam (2026a) Anti-ARPC5/p16 ARC antibody [EP1551Y] (ab51243). Available at: Abcam product page (Accessed: 28 July 2026). (Abcam)
Abcam (2026b) Anti-ARPC5/p16 ARC antibody (ab151303). Available at: Abcam product page (Accessed: 28 July 2026). (Abcam)
David, S. (2026) ‘Mind over antibody’, For Better Science, 2 June. Available at: For Better Science (Accessed: 28 July 2026). (For Better Science)
Mallapaty, S. (2026) ‘Protein name confusion created antibody mix-up affecting hundreds of papers’, Science, 5 June. Available at: Science (Accessed: 28 July 2026). (Science)
Research Resource Identification Initiative (RRID) (2026) Anti-ARPC5/p16 ARC antibody [EP1551Y], RRID: AB_2059963. Available at: RRID Resolver (Accessed: 28 July 2026). (rrid.site)
UniProt Consortium (2026a) CDKN2A: Cyclin-dependent kinase inhibitor 2A, Homo sapiens. UniProtKB accession P42771. Available at: UniProt (Accessed: 28 July 2026). (UniProt)
UniProt Consortium (2026b) ARPC5: Actin-related protein 2/3 complex subunit 5, Homo sapiens. UniProtKB accession O15511. Available through the target record referenced in the Abcam product documentation (Accessed: 28 July 2026). (Abcam)
