How to reproduce an ELISA from a paper

This page is about reproducing a published ELISA measurement, including the plate layout and the standard curve that turn optical density into a concentration. Papers commonly report that a commercial kit was used according to the manufacturer's instructions, which hides sample dilution, curve fitting, and whether readings fell inside the quantifiable range. Those choices decide whether your numbers can be compared with theirs.

What methods sections usually leave out

Sample dilution factor and the diluent used
Concentration is read off the curve and then multiplied by the dilution. If the dilution is unstated, a reported value cannot be reconstructed, and matrix effects differ between diluents.
Standard curve range, number of points, and the fitting model
A four-parameter logistic fit and a linear fit over a narrow range give different values for the same optical density. The lowest and highest standards also define what counts as quantifiable.
Whether samples fell inside the quantifiable range
Readings above the top standard or below the lowest are extrapolations. Papers often report a mean without saying whether some samples were out of range.
Plate layout and replicate structure
Where standards, samples, and blanks sit on the plate affects results through position effects, and duplicate wells are not biological replicates.
Sample handling before the assay
Freeze-thaw cycles, storage temperature, whether serum or plasma was used, and the anticoagulant can all change measured concentrations of a given analyte.
Incubation times, temperatures, and wash count for a custom assay
For an in-house ELISA, coating concentration, blocking, and the wash protocol drive both signal and background, and these are the first details to disappear from a methods section.

Key parameters and what they change

Parameter	Typical	Changes together with
Coating antibody or antigen concentration	Titrated for in-house assays; fixed and undisclosed in most commercial kits	Maximum signal and background. Raising the coating concentration shifts the usable range of the curve.
Detection antibody dilution	Set by titration against the coating and the standard range	Curve slope and the optical density ceiling, and the volume of diluted antibody needed per plate.
Standard series dilution factor	Commonly a two-fold or three-fold serial dilution across the standards	Volumes per dilution tube and the total volume of top standard needed; it also sets curve resolution.
Sample dilution	Chosen so expected concentrations fall in the middle of the curve	Volume of sample and diluent per well, and the back-calculated final concentration.
Wells used per plate	Samples times replicates, plus standards, blanks, and controls	Every reagent total: coating, blocking, antibody, substrate, and wash buffer.
Substrate development time	A fixed interval before stopping, per the assay	Optical density of all wells. Stopping late can push the top standards into saturation.

Troubleshooting

Symptom	Likely causes	What to check first
Standard curve flattens at the top	Signal saturation from long development, too much detection antibody, or a top standard above the assay's dynamic range.	Shorten development, dilute the detection antibody, and verify that samples sit on the responsive part of the curve rather than the plateau.
High coefficient of variation between replicate wells	Pipetting error with small volumes, incomplete mixing of the dilution series, bubbles, or inconsistent washing.	Pipette larger volumes where possible, mix each dilution step thoroughly, inspect for bubbles before reading, and check that all wells are aspirated equally.
Outer wells read systematically higher or lower	Edge effect from evaporation and temperature gradients during incubation.	Equilibrate plates to temperature before starting, seal during incubations, and distribute standards and samples so that no condition sits only at the edge.
High background in blank wells	Insufficient blocking, carryover from inadequate washing, or contaminated substrate.	Add washes, extend blocking, and test a fresh substrate aliquot in a few blank wells.
Sample values differ from the paper by a constant factor	A different dilution factor, a different kit or standard calibration, or different units.	Recompute from the raw optical density with the paper's stated dilution and units, and confirm which kit, standard, and fitting model were used.

Worked example: from a methods sentence to executable steps

Analyte concentrations in culture supernatants were measured by sandwich ELISA according to the manufacturer's instructions. Absorbance was read at 450 nm and concentrations were determined from a standard curve.

Each step is tagged by how clearly the text states it: explicit, partial, inferred, or missing.

1.
Collect and clarify culture supernatantsPartial
Collection is implied; centrifugation and storage conditions are not stated.
2.
Prepare the standard series by serial dilutionPartial
Range, number of points, and dilution factor come from the kit insert, not the paper.
3.
Dilute samples in assay diluentMissing
No dilution factor given, so reported concentrations cannot be back-calculated.
4.
Add standards and samples to the coated plate and incubateInferred
Incubation time and temperature follow the kit insert; the paper does not confirm them.
5.
Wash, add detection antibody, and incubateInferred
Follows the insert; record the version you used.
6.
Add substrate, develop, and stop the reactionInferred
Implied by the 450 nm read; development time is not stated and strongly affects the top of the curve.
7.
Read absorbance at 450 nmExplicit
8.
Fit the standard curve and calculate concentrationsPartial
A curve is mentioned but not the fitting model or whether samples were in range.

How Vara helps

Paste the methods text or upload the kit insert as a PDF or photo, and Vara turns it into plate setup, incubation, and readout steps with extracted parameters.
Fields like sample dilution, curve range, and fitting model are tagged explicit, partial, inferred, or missing, and the missing ones are listed as reproduction risks.
Plate layout, standard series, and reagent volumes are linked by formulas, so changing the sample count or plate format recalculates the dilution series and totals.
Each plate run is logged against the protocol version, so a shift in the standard curve can be traced to a reagent lot or a change in development time.

Try Vara on your own protocol See how importing works

FAQ

What dilution should I use when the paper does not say?+

Run a small pilot with two or three dilutions spanning the expected range, then use the one that places your samples near the middle of the standard curve. Record the factor, because the final concentration depends on it directly.

Why is my ELISA standard curve not linear?+

Sandwich immunoassay curves are generally sigmoidal, not linear, when plotted against log concentration, so a four-parameter logistic fit is usually appropriate and only the central region is approximately linear. If the top of the curve is flat, the signal is likely saturated, which can be reduced by shortening substrate development or diluting the detection antibody.

What causes high well-to-well variation in an ELISA?+

Most often pipetting of small volumes, incomplete mixing in the dilution series, bubbles in the wells, and uneven washing. Evaporation at the plate edges adds a positional component. Sealing plates and distributing conditions across the plate address the latter.

Can I compare my ELISA values directly with the published numbers?+

Only with care. Absolute concentrations depend on the kit, the standard material, the sample matrix, and the dilution, so values from different kits often differ systematically. Comparing the direction and size of differences between conditions is more robust than comparing absolute values.