Comparing ELISA and western blot evidence

Comparing ELISA and western blot evidence is how you stop a single antibody picture from becoming the whole claim. An ELISA, when the standard curve and the controls hold, reports a number or a rank in a well. A western blot reports a band at an apparent mass on a membrane. They share antibodies and they do not share a sample history. One can stay native enough to see a complex. The other usually denatures. Used together, they disagree in ways that are informative. Used alone, each one hides the failure mode the other would have shown. Plate formats and the meaning of a curve are in ELISA formats, controls and readout. Transfer, blocking and molecular weight ladders are in western blot from gel to membrane. The membrane in the photograph is the blot half of this comparison. Neither half is a diagnostic test.

Two claims, and the decision they support

The decision is whether a change you care about survives contact with a second physical method. If you only need a calibrated rank in a liquid matrix, and the curve, the blank and the spike recovery behave, the ELISA may be the assay you report. If you need to know that the antibody is painting a protein of the expected size, the blot is doing work the plate cannot do, because a well has no molecular weight. If the two disagree, you do not average them. You identify which assumption broke.

Write both claims before you look at a combined figure. The ELISA claim names the analyte, the unit, the matrix and the direction of change. The blot claim names the apparent mass, the loading control and the antibody lot. A figure that shows a bar and a lane without those sentences invites a reader to treat them as the same evidence. They are corroboration only where their assumptions overlap.

What the number in the well actually required

A sandwich ELISA builds a complex of capture antibody, analyte and detection antibody. The analyte must present two epitopes at once, or a repeated epitope that both reagents can share without blocking each other. The matrix stays comparatively native: serum, medium, or a lysate that was never boiled in a strong detergent. The standard curve translates optical density into the unit of the calibrator, and only inside the range of the standards. Nonspecific binding is controlled by the blank and by wells that omit one reagent. If those wells are loud, you do not have a number yet.

That architecture is blind to mass. A cross-reactive protein that carries the epitopes will be counted as analyte. A fragment that carries both epitopes will be counted. A complex that displays the epitopes will be counted even if the polypeptide you name on the axis is only one subunit. The curve does not know. Sequence context from UniProt tells you which domains exist. It does not tell you which ones survived in the well.

What the band required

A typical immunoblot runs proteins through sodium dodecyl sulphate, often after reduction of disulphide bonds, transfers them to a membrane, and probes with antibody. The photograph of forceps and a buffer tray is that object. Apparent mass comes from how the band sits relative to a ladder, and post-translational modifications, incomplete reduction and unusual amino acid composition all move bands. The method is explained further in the blot article. The point here is the sample treatment: the protein is unfolded, and many conformational epitopes are gone.

A band at the wrong mass is positive evidence that the antibody binds something else, or that the protein does not migrate where you hoped. A missing band is weaker evidence. Transfer can fail, the epitope can be denaturation-sensitive, and the load can be below what the blot can show. A missing band does not by itself refute an ELISA that was reading a native complex. A present band at an unexpected mass does force you to stop calling the ELISA specific until you explain the band.

Loading controls and total protein stains stop a dark band from being a dark lane. They do not create a standard curve. Unless you loaded known masses of a calibrator, the blot remains comparative. NCBI Bookshelf is a public molecular biology reference you can use to recall how denaturing gels separate polypeptides. It is not a protocol for your antibody.

Where agreement is meaningful

Agreement means the ELISA changes in the same direction as the specific band, the blank and the loading control behave, and a negative sample is quiet in both methods. That pattern supports a claim that the epitope the two methods share has changed in abundance or in availability, under the sample preparations you used. It does not prove biological activity. It does not identify which isoform if both methods would see several.

Run the comparison on split aliquots of the same specimen when you can. An ELISA on serum and a blot on a cell pellet from a different day are two experiments. They may both be true and still refuse to match. Say that in the record so a reader does not invent a contradiction you did not test.

Two disagreements that earn their keep

First pattern: the ELISA moves, the blot does not. Suspect the epitope and the denaturation. The sandwich may require a folded surface that the blot destroyed, so the blot is silent for a reason that has nothing to do with abundance. The ELISA may also be seeing a complex, a shed ectodomain, or a cross-reactive protein that never formed a band at the mass you cropped. Look at the full blot, not a cut lane. Check whether the supplier claims the antibody for denatured applications. If the claim is native only, the blot was the wrong orthogonal test, and you need a native method or a second ELISA antibody.

Second pattern: the blot moves, the ELISA does not. Suspect the matrix and the hook effect. Neat serum can sit on the plateau of a curve, so a real change in the blot's cleaned lysate does not move the well. Excess analyte can also create a hook, where the signal falls as the sample gets more concentrated, because capture and detection are occupied separately and the sandwich never forms. Dilute the ELISA samples across a series. If a diluted well reads higher than the neat well, you were past the assay. If dilutions are flat and the spike recovery is poor, the matrix is hiding the analyte and the blot, which used a different extraction, was free to move.

A third, less tidy pattern is a band at the wrong mass plus a lively ELISA. Do not average that into a success. The antibody binds more than one thing, or the mass assignment is wrong. Identify the extra band, change antibody, or narrow the claim to what both methods can support.

How to move when they disagree

Follow the branch you actually have, not both at once. If the ELISA is the one that moved, confirm the curve, then ask whether the blot epitope can survive denaturation. A native gel or a different antibody is a fairer second test than repeating the same denaturing blot and hoping. If the blot is the one that moved, dilute the ELISA and check the hook before you question the membrane. Spike a known amount into the ELISA matrix and see whether you recover it. Poor recovery means the well was never going to track the lysate.

Transfer quality still belongs to the blot article: a lane that failed to transfer should not be called a negative ELISA confirmation. Ponceau or another total protein check, and the ladder, stay in the file. Worked plate and blot layouts on protocols.io are reminders of what to record, not times to copy onto a different antibody.

Safety and research use

Blot membranes, methanol in some transfer buffers, and ELISA stop acids are chemical hazards with their own sheets. Antibody stocks may contain preservatives. Human and animal specimens follow institutional biosafety rules. The WHO Laboratory Biosafety Manual is a public reference for that institutional decision. Agreement between a plate and a membrane does not convert either method into a diagnostic report. Say what the research comparison supports, and stop there.

Heat on the bench and a membrane that dries

A colorimetric ELISA developed in a hot room cannot be compared with one developed in a cooled room as if the clocks matched. A membrane that dries out while you wait for an imager in a dry, air-conditioned room can leave uneven background that looks like extra bands. Keep the membrane wet in the way the detection protocol asks, and do not crop a tide mark into a figure. If a power cut stops the transfer halfway, that blot is not a partner for the ELISA you finished in the morning. Record the failed transfer and repeat it. Do not explain a biological disagreement with a half-transferred gel.

What to put in an enquiry

Name the analyte, whether you need an ELISA pair, a blot-capable antibody, or both, and whether the epitope must survive denaturation. State the matrix for the plate and the lysate type for the blot. The reagents and chemicals catalogue is a starting point for reagent classes. The molecular biology pathway is the wider context. Use the quote request to ask whether a quotation is possible. A method can be discussed from the two claims you are trying to align. Ask the reply to state application claims for ELISA and for blot separately. Do not accept one photograph as cover for the other method.