Before you trust any antibody, confirm application-specific validation, independent specificity evidence (knockout, orthogonal, independent-antibody or capture mass spectrometry data), clonality or recombinant status, species reactivity, and immunogen or epitope information, and check that the supplier will support you. This antibody selection guide shows how to read that evidence and run pragmatic checks in your own lab.
TL;DR:
- Treat vendor application claims as hypotheses: prioritize data from your assay and comparable samples, especially endogenous expression, and verify with positive and negative controls.
- Choose recombinant or monoclonal antibodies for repeat studies and long term supply, but consider polyclonals for scarce targets or damaged, masked epitopes.
- Match validation to sample treatment and species; predicted reactivity rests on sequence similarity, so pilot less common models and check epitope level conservation.
- For specificity, combine independent evidence where feasible: knockout controls, orthogonal measurements, or antibodies against different epitopes each reveal different limitations.
- Before testing, define acceptance criteria, titrate the antibody, replicate key comparisons, and retain a validated reference sample for future lot checks.
Table of Contents
- How do you shortlist antibodies before any lab work?
- Which antibody type suits your study: monoclonal, polyclonal or recombinant?
- What does strong validation evidence look like, and how should you read it?
- How does application compatibility change what validation you need?
- Why do species reactivity and host species matter?
- What should you learn from the immunogen and epitope information?
- How do fixation and sample processing affect antibody performance?
- Should you use direct detection or secondary antibodies?
- How do you validate a new antibody in your own lab?
- What should you do when an antibody underperforms?
- How do affinity and sensitivity affect antibody choice?
- Why does antibody isotype matter for assay design?
- How should you store and handle antibodies?
- What artefacts do antibodies produce, and how can you recognise them?
- Are recombinant antibodies more reproducible than traditional ones?
- How should you budget for antibody selection?
- Why open validation data matters most, in our view
- Find validated antibodies faster with our catalogue and ABMIUM MATCH
- FAQ
- Sources
How do you shortlist antibodies before any lab work?
A fast shortlist saves more bench time than any troubleshooting later. Work through the steps below in order, and discard any candidate that fails an early step instead of rescuing it with extra optimisation.
- Define the experiment precisely. Write down the assay, the species, the sample type, how the sample is treated (lysed, fixed, embedded, live) and the endpoint you will measure. An antibody is never “good” in general, only good for a defined setup.
- Demand application-specific validation first. Evidence generated in your assay, on a comparable sample, outweighs a long list of applications on a datasheet. Where it is missing, require several independent lines of supporting evidence.
- Count the validation pillars. Check which specificity strategies the supplier actually shows, not merely claims, and whether the data are on view.
- Confirm species reactivity and host compatibility. Check that the species you work with is covered by tested data, and that the host species suits your secondary reagents and any co-staining panel.
- Record clonality, clone number and supply continuity. Recombinant and monoclonal reagents with a named clone offer the strongest safeguard against drift between lots.
- Inspect the immunogen and the raw images. A described immunogen and uncropped blots or images let you judge the evidence yourself.
- Prefer suppliers with open data and technical support. Someone who can answer a specific question about a lot is worth more than another page of claims.
Guidance on selecting antibodies follows the same logic: compare vendors, match antibody type to application, prefer complete validation data, and validate the promising candidates yourself against acceptance criteria you set beforehand. Our own antibody specificity testing checklist expands step three into concrete tests.
Which antibody type suits your study: monoclonal, polyclonal or recombinant?
The production route determines how consistent a reagent will be across years, and it shapes sensitivity and specificity too.
- Monoclonal antibodies come from a single hybridoma clone and recognise one epitope. They are consistent within a clone, though hybridoma drift or cell line loss can still threaten long-term supply.
- Polyclonal antibodies are purified from the serum of an immunised animal and contain a mixture of antibodies against several epitopes. That breadth often raises signal for low-abundance targets and tolerates mild epitope damage, but every new animal and bleed is effectively a new reagent.
- Recombinant antibodies are produced from a defined sequence expressed in a host cell line. Because the sequence is fixed, the reagent can be remade identically, which removes most batch-to-batch concern.
In practice, favour recombinant or monoclonal reagents for longitudinal studies, multi-site work and anything heading towards regulated use, where you may need to buy the same reagent again later. Polyclonals remain a reasonable choice for detecting scarce proteins, for applications where a denatured or partly masked epitope is likely, and for early exploratory work. Record the clone identifier, or the recombinant sequence where it is disclosed, in your lab notebook and methods section, because that identifier is what keeps a future repeat comparable. Community consensus protocols such as those used by YCharOS likewise prioritise recombinant or monoclonal antibodies with clone numbers. If continuity matters to you, our recombinant antibodies collection is a sensible place to start.
Pro Tip: Store the clone number, supplier catalogue number and lot number together in one record for every experiment, so a failed repeat can be traced to the reagent rather than guessed at.
What does strong validation evidence look like, and how should you read it?
The International Working Group for Antibody Validation (IWGAV) set out five conceptual pillars for assessing antibody specificity, and recommends using as many as feasible to improve reproducibility. The same proposal is summarised in the Human Protein Atlas news archive.
- Genetic strategies: knockout or knockdown controls remove or reduce the target, so a specific antibody should lose its signal. Their limit is that a knockout cell line must be available and the target must be expressed in it to begin with.
- Orthogonal strategies: antibody signal is compared with an antibody-independent measurement of the same target, such as mass spectrometry or transcript levels, across samples with differing abundance. They work best when expression varies widely between samples.
- Independent antibody strategies: two or more antibodies raised against different epitopes of the same target should give concordant results. The weakness is that concordant antibodies can still share an off-target.
- Recombinant expression: a tagged version of the target is expressed and the antibody signal is compared with tag detection. This is informative but sits some distance from endogenous biology.
- Immunocapture mass spectrometry: the antibody pulls down its target, and mass spectrometry identifies what was captured, which reveals both the intended target and any interacting or cross-reacting proteins.
No pillar is sufficient alone, and each performs differently by assay. An enhanced validation programme adapted the pillars for western blot and validated thousands of antibodies, while stressing that each pillar has limits and that validation must be application-specific.
Vendor application recommendations are not a guarantee of success: the burden of validation is shared, and researchers should independently verify performance with positive and negative controls suited to their own experimental setup. IWGAV guidance
From vendor data, ask for full uncropped blots, more than one sample type, the method details, and the number of pillars genuinely demonstrated. Prefer data generated at endogenous expression, because overexpression can mask cross-reactivity and give false confidence in specificity. For a practical rendering of these pillars at the bench, see our antibody validation methods guide.
How does application compatibility change what validation you need?
Validation in one assay does not transfer to another, because the form in which the epitope reaches the antibody differs. Western blot presents denatured, linearised protein, whereas IHC, immunofluorescence, flow cytometry and sandwich ELISA generally present protein in a more native or fixed context. An antibody that sees a denatured epitope on a membrane may never reach the same epitope folded inside a fixed cell.
A common vendor pitfall is relying on purified or overexpressed protein as the only control. That demonstrates the antibody can bind its target, but says little about whether it avoids other proteins in a complex sample.
- Western blot: look for a single band at the expected size in a lysate from relevant cells, a loss of signal in a knockout or knockdown control, and a loading control. Our western blot antibody selection checks walk through this in detail.
- IHC and immunofluorescence: look for staining in the expected subcellular compartment, concordance with a second antibody or with transcript data, and a negative tissue or knockout control processed identically.
- Flow cytometry: look for a clear separation between positive and negative populations, an isotype or knockout control, and information on whether surface or intracellular staining was used.
- ELISA: look for a defined matched pair, a standard curve in a relevant matrix, and spike-and-recovery or dilution linearity data for your sample type.
Whatever the assay, ask one question of the evidence: does it come from a sample and a treatment comparable to mine? If not, treat the application claim as a hypothesis to test rather than a result to rely on.
Why do species reactivity and host species matter?
Species reactivity tables list the species in which an antibody has been tested or predicted to work, and the difference between those two words matters. A “predicted” entry usually rests on sequence similarity between the immunogen and the orthologous protein, while a “tested” entry rests on data. Higher sequence identity across the epitope raises the likelihood of cross-reactivity, but identity elsewhere in the protein tells you little.
- Check reactivity at the epitope level. Align the immunogen sequence with your species’ protein using reference records such as NCBI Gene and its linked sequence databases.
- Match host to your secondary system. The host species of the primary determines which secondary antibody you can use, and a mouse primary on mouse tissue will produce high background from anti-mouse secondaries binding endogenous immunoglobulin.
- Account for isotype and subclass. In multiplexed staining, two primaries of the same host need different isotypes or subclasses, or direct conjugation, to be told apart.
- Ask for cross-reactivity data. Suppliers often hold unpublished data from additional species, and asking is free.
- Run a pilot. Before committing a precious sample, test the antibody on a small positive and negative pair from your species.
If you work in a less common model organism, assume nothing: treat a predicted reactivity claim as unverified until your own pilot gives a clean result.
What should you learn from the immunogen and epitope information?
The immunogen is the molecule used to raise the antibody, and its description is one of the most informative lines on a datasheet. Common descriptors include a short synthetic peptide, a recombinant protein fragment, a protein epitope signature tag (PrEST), or the full-length protein.
- Peptide immunogens target a defined short region, which makes epitope location clear but can yield antibodies sensitive to folding or modification.
- Recombinant fragments and PrESTs present a longer stretch of sequence and tend to produce antibodies that recognise more than one conformation.
- Full-length protein can raise antibodies against many epitopes, which helps sensitivity but widens the opportunity for cross-reactivity with related family members.
Epitope location matters because it decides whether the antibody works after a given sample treatment and in a given species. An epitope buried in a protein-protein interface may disappear in native samples, and one in a region that varies between species may fail on your orthologue. Likewise, an epitope within a conserved domain shared by paralogues raises the odds of off-target binding.
If the immunogen is not disclosed, request it, along with the raw validation images. Where the supplier cannot or will not provide them, choose an alternative whose provenance is clearer.
Pro Tip: Align the immunogen sequence against the paralogues of your target as well as the orthologues in your species; a high-identity paralogue is the most common hidden source of unexpected bands.
How do fixation and sample processing affect antibody performance?
An antibody validated on fresh, unfixed lysate can fail completely on formalin-fixed, paraffin-embedded (FFPE) tissue, and the reason is almost always epitope accessibility. Fixation cross-links proteins and can mask or alter the epitope, while denaturing preparations expose epitopes that stay hidden in a native fold. The reverse also happens: a conformation-dependent antibody may work on live cells yet give nothing on a reduced, denatured blot.
- Match the validation to your treatment. Ask whether data exist for your fixative, fixation time and embedding method, not just for “IHC” in general.
- Plan antigen retrieval. Heat-induced or enzymatic retrieval can restore masked epitopes in fixed tissue, but the best buffer and pH vary by antibody, so test a small range rather than assuming one condition.
- Use matched controls. Include a tissue or cell pellet known to express the target and one known not to, processed in the same batch and with the same retrieval.
- Watch for over-fixation. Prolonged fixation increases masking and background, so keep fixation times consistent across a cohort.
- Pilot unusual treatments. For decalcified tissue, cleared samples, cryosections or unusual detergents, run a small pilot series before scaling up.
When you must work with archival FFPE material, which cannot be re-collected, spend the extra time on a retrieval titration. It is far cheaper than discovering after the full cohort has been stained that the epitope was never accessible.
Should you use direct detection or secondary antibodies?
Directly conjugated primary antibodies shorten the protocol, remove secondary cross-reactivity and simplify multiplexing, but they usually give less amplification than indirect detection and each target needs its own conjugate. Indirect detection with a labelled secondary adds signal because several secondary antibodies can bind one primary, and it lets you swap detection chemistries without changing the primary.
- Match the secondary to the primary’s host and isotype. An anti-rabbit secondary is for rabbit primaries, and for mouse primaries you may need subclass-specific secondaries to separate co-stains.
- Use cross-adsorbed secondaries for multiplexing. These have been depleted against immunoglobulins of other species, which reduces unwanted binding between detection layers.
- Watch Fc binding. Tissues and cells such as macrophages and B cells carry Fc receptors, so block appropriately or use fragment-based secondaries.
- Choose fluorophores by instrument and brightness. Pair the brightest fluorophores with the least abundant targets, and check spectral overlap on your own filter set or cytometer configuration.
- Choose enzymatic labels by readout. Horseradish peroxidase and alkaline phosphatase systems suit chemiluminescent or chromogenic detection, and the substrate you pick affects sensitivity and dynamic range.
Always run a secondary-only control. It is the quickest way to establish how much of your background comes from the detection layer rather than the primary.
How do you validate a new antibody in your own lab?
Even a well-documented antibody should earn your trust in your hands. A minimal workflow, adapted to what is feasible, looks like this:
- Run positive and negative controls. Use a sample known to express the target and one that does not, processed identically.
- Add a genetic control where available. A knockout or knockdown line gives the clearest test of specificity, and our knockout validation timeline shows how to plan it.
- Compare orthogonally. Check that the antibody signal tracks an independent measure of the target across samples with different abundance.
- Compare against an independent antibody. Our independent antibody validation protocol describes when this adds real information.
- Titrate. Test a dilution series to find the working range rather than adopting the datasheet dilution.
- Replicate and check lot-to-lot. Repeat on independent days, and when you buy a new lot, compare it against the retained old one on the same samples.
- Write acceptance criteria first. Decide in advance what counts as a pass, such as a single band at the expected size that disappears in the knockout, so the result cannot be rationalised afterwards.
As a rule of thumb, independent replicates for each key comparison give you a sense of variability, though your assay’s noise should guide how many to run. A pragmatic minimal workflow, covering controls, titration and one orthogonal or independent comparison, is usually achievable in a few weeks of intermittent bench time, with longer timelines for knockout generation. Where in-house capacity is limited, specialist collaborators can run the work for you.
Pro Tip: Aliquot and freeze a reference sample from your first successful validation run, then include it in every later experiment as a standing positive control for lot checks.
What should you do when an antibody underperforms?
Before blaming the reagent, rule out the common mundane causes. Most failures are fixed in a handful of quick checks.
- Concentration: titrate the antibody, because too much drives background and too little loses signal.
- Incubation: adjust time and temperature, trying a longer incubation at a cool temperature for weak signals.
- Blocking: switch blocking agent, since milk can interfere with some phospho-specific and biotin-based detection.
- Washing: increase the stringency and number of washes if background is high.
- Sample integrity: check for protein degradation, freeze-thaw damage or poor fixation.
- Unexpected bands or staining: these may be isoforms, post-translational modifications, degradation products or cross-reactivity. Rule out the last by testing a knockout or knockdown control, or a second antibody against a different epitope.
If none of this helps, escalate. Contact the supplier’s technical team with your protocol and images, and ask about return or replacement terms. Where the stakes are high, independent validation by a specialist can settle the question objectively.
How do affinity and sensitivity affect antibody choice?
Affinity describes how tightly one binding site holds its epitope, whereas sensitivity is a property of the whole assay: the lowest amount of target you can reliably detect. The two are related but not interchangeable, since sensitivity also depends on target abundance, epitope accessibility, detection chemistry and background.
High affinity helps in applications with stringent washes or short incubations, and for capturing scarce targets in immunoprecipitation or sandwich ELISA. It is less decisive in a western blot where you can compensate with longer incubation or stronger amplification. Very high affinity can even be a drawback when you want to elute a target gently under native conditions.
Suppliers rarely publish quantitative affinity values, so judge sensitivity empirically. Run a dilution series of a positive sample, and note the lowest input at which you still obtain a clean, quantifiable signal. For low-abundance targets, polyclonals may add signal by binding several epitopes at once, while a high-quality recombinant monoclonal gives the most reproducible result across repeated experiments. Favour specificity over raw signal: a bright, off-target signal is worse than a modest, correct one.
Why does antibody isotype matter for assay design?
The isotype (for example IgG1, IgG2a or IgM in mouse) determines how the antibody behaves with secondary reagents, Fc receptors and certain purification methods. It is a design variable, not a footnote on the datasheet.
In multiplexed staining, two mouse primaries of different IgG subclasses can be detected with subclass-specific secondaries, avoiding the need for different host species. For flow cytometry, a matched isotype control helps estimate nonspecific binding, though it is a weak substitute for a true negative sample or knockout. In cells carrying Fc receptors, an intact IgG can bind through its Fc region regardless of antigen specificity, which is why blocking or Fab and F(ab’)2 fragments are used.
Isotype also affects behaviour in immunoprecipitation, where Protein A and Protein G bind different species and subclasses with differing strength, so check compatibility with your capture resin. IgM antibodies are larger, can be more avid but harder to purify, and are not generally used as ordinary detection reagents in blotting. Note the isotype alongside clone number when you record a reagent, because it constrains every downstream choice.
How should you store and handle antibodies?
Antibodies are proteins, and poor handling erodes performance before the experiment even begins. Follow the supplier’s storage instructions for the specific product, as they take precedence over general advice, and apply these habits across the board.
- Avoid repeated freeze-thaw cycles. Aliquot on first thaw into volumes you will use in one go, and store aliquots as directed.
- Keep working stock cold. Store refrigerated stock at the recommended temperature and return it promptly after use.
- Protect conjugated reagents from light. Fluorophore conjugates photobleach, so store them in the dark and minimise exposure at the bench.
- Mind the buffer. Additives such as glycerol or stabilising protein affect storage temperature, and some preservatives, such as sodium azide, interfere with enzymatic detection and live-cell work.
- Prevent contamination. Use clean tips, and do not return unused diluted antibody to the stock tube.
- Label thoroughly. Record catalogue number, lot, date of receipt and date of aliquoting.
Diluted working solutions generally lose stability faster than concentrated stocks, so prepare them fresh unless the supplier states otherwise. If performance drifts over time, test a freshly thawed aliquot alongside the old one before concluding the reagent has failed.
What artefacts do antibodies produce, and how can you recognise them?
Many apparent biological findings are artefacts of the reagent. Learning their signatures saves months of chasing false leads.
- Extra bands on a western blot: these may reflect cleavage products, isoforms, modifications or off-target binding. A band that persists in a knockout sample is an off-target signal.
- Smears and high background: often caused by excessive antibody, insufficient blocking, or sample overload.
- Diffuse or uniform staining: frequently nonspecific binding of the antibody or secondary, which a secondary-only control will reveal.
- Edge effects and patchy staining in tissue: commonly drying during incubation, uneven reagent coverage or inconsistent retrieval.
- Nuclear or Fc-mediated staining in the wrong compartment: can result from Fc receptor binding or from charged, sticky antibodies.
- Signal that vanishes between lots: points to lot variability, more likely with polyclonals, and is exactly why you should retain and compare an earlier lot.
- Cross-channel bleed-through in multiplexing: reflects spectral overlap or secondary cross-reactivity rather than co-localisation.
The most reliable defence is the control set you already planned: a negative sample, a genetic control where possible, and an omission of the primary. Treat any result you cannot reproduce with an independent antibody as unconfirmed.
Are recombinant antibodies more reproducible than traditional ones?
Recombinant antibodies are expressed from a defined sequence, so the reagent can be remade to the same specification whenever it is needed. Hybridoma-derived monoclonals depend on a living cell line that can drift, mutate or be lost, and polyclonals depend on individual animals whose immune response differs from bleed to bleed and cannot be repeated exactly.
For reproducibility, the practical ranking is therefore recombinant first, then monoclonal, then polyclonal, with the caveat that a well-characterised polyclonal can still outperform a poorly characterised recombinant. Sequence-defined reagents also allow engineering, for example switching the host species or isotype to suit a multiplex panel while keeping the binding domain constant, and they avoid animal use in routine production.
Reproducibility is not the same as specificity, though. A recombinant antibody that binds the wrong protein is reproducibly wrong, so the validation requirements described earlier apply equally. Community knockout-based characterisation work, described in a consensus platform paper, compares antibodies of every type across western blot, immunoprecipitation and immunofluorescence using isogenic knockout controls, which is the kind of head-to-head evidence worth looking for.
How should you budget for antibody selection?
The purchase price is the smallest part of the real cost of an antibody. A cheap reagent that fails costs you the sample, the consumables, the weeks of repeat work and sometimes a delayed paper. Budget for the whole path to a trustworthy result.
- Plan for validation. Reserve a fraction of the project budget and time for validating each critical reagent in your hands, especially anything underpinning a central claim.
- Compare cost per usable experiment. A recombinant or well-validated antibody that works at high dilution may cost less per assay than a cheaper one needing more material and repeat runs.
- Buy enough of one lot. For a long study, purchasing a sufficient quantity of a single lot, after it passes validation, protects consistency.
- Value the evidence. Open validation data, technical support and a clear return policy all have monetary worth.
- Use institutional pricing. Teams buying regularly may qualify for contract pricing, which we offer to institutions.
- Reserve for contingency. Include a second candidate antibody in the plan so a failure does not stall the project.
Current prices are on our website, and we are happy to compare options with you on cost per validated result rather than catalogue price alone.
Why open validation data matters most, in our view
Across the failures we see researchers describe, the pattern is consistent: the reagent was chosen from a datasheet, the datasheet claimed more than the data showed, and the gap only appeared after samples were spent. The fix is not more optimisation but earlier scrutiny of evidence. Open validation data, with full images, named methods and clear provenance, turns an antibody purchase from a bet into a checkable decision.
That is why we think the strongest signal of a reliable reagent is a supplier willing to show its work, including the experiments that did not go perfectly. It is also why we offer consultative sourcing, in which a scientist helps compare candidates against your actual sample and assay. Independent validation, run by people with no stake in the answer, removes a further layer of doubt. For deeper reading on any step here, our research validation blog covers the methods and timelines in practical detail.
— Veron
Find validated antibodies faster with our catalogue and ABMIUM MATCH
We built our catalogue around the checks described in this guide. In our primary antibodies collection you can search and filter by target, application, species, host, clonality and available validation evidence, so the shortlist step can be quicker. Products come from visible manufacturers, which keeps provenance clear.
If your experiment is unusual, or the evidence is hard to interpret, ABMIUM MATCH is our scientist-led sourcing service. You send us your experimental requirements, and we help you compare suitable products and the evidence behind them. We can also arrange independent product validation, native mass spectrometry and spatial validation through specialist collaborators, and where no suitable reagent exists we offer custom antibody, protein and ELISA development. Teams buying at volume can ask about institutional pricing.
To start, browse the primary antibodies collection, or contact us through ABMIUM MATCH and tell us your assay, species and sample type. We will help you compare reagents before you spend a single sample.
FAQ
How many validation pillars should an antibody have?
There is no fixed number, but more independent lines of evidence give stronger support. The International Working Group for Antibody Validation advises using as many of its five pillars as feasible. What matters most is that the evidence matches your application.
Are recombinant antibodies better than monoclonal or polyclonal ones?
For consistency over time, usually yes, because a defined sequence can be reproduced exactly. They are not automatically more specific, so they still need validation in your assay. Polyclonals can remain useful for scarce targets or exploratory work.
Is a vendor’s recommended application enough to trust an antibody?
No. An application listed on a datasheet is not a guarantee that the antibody will work in your hands. Guidance holds that vendor recommendations do not remove the need for you to verify performance with suitable positive and negative controls.
Why should validation use endogenous protein rather than overexpression?
Overexpression can hide cross-reactivity, because the target is present at an unnaturally high level that swamps off-target signals. Validation at endogenous levels is a closer test of real performance, which is why practitioners prioritise it.
What should I do if the immunogen is not listed?
Ask the supplier for the immunogen sequence and the raw validation images. If they cannot supply them, choose an alternative with clearer provenance, since you cannot predict epitope location or cross-reactivity without that information.