How to use (3 steps)
- Select an example or paste standards (concentration and signal).
- Choose 4PL/5PL, weighting, and other settings.
- The curve and unknown concentrations appear (extrapolation is flagged).
This is a model. Extrapolation outside the standard range can be inaccurate. Use blank subtraction only when needed.
Need help with blank handling, 4PL vs 5PL, or extrapolation?
Use this calculator for the actual ELISA fit, weighting trials, recovery checks, and unknown concentration estimates. Start with 4PL, inspect residuals and recovery, then try 5PL only when the standard curve shows clear asymmetry that the simpler model cannot explain.
Open the assay standard curves guide when you need the reasoning behind blank treatment, model choice, range checks, and report wording.
Inputs (standards, unknowns, settings)
—
Display options
Minimum columns: concentration and signal. Multiple rows with the same concentration are treated as replicates.
Minimum columns: sample and signal. If dilution_factor is blank, it is treated as 1.
Results (curve & concentrations)
R² and AIC are guides. Also check residuals and whether values are extrapolated.
Standard curve
Residual plot (optional)
Parameters (A, B, C, D, G)
| param | value | 95% CI (approx) |
|---|
Guide: A=Bottom (low concentration), D=Top (high), C=EC50 (midpoint), B=Hill slope, G=asymmetry (5PL only). Internally we compute (x/C)^B as exp(B·ln(x/C)). Log10 on x-axis is for readability.
QC (back-calculation / recovery)
| label | conc_nominal | signal_mean | conc_hat | recovery |
|---|
Unknowns (y → concentration)
| sample | signal_mean | DF | conc_measured | conc_original | flag |
|---|
How it’s calculated
- Fit a 4PL/5PL logistic curve with nonlinear least squares.
- Unknown concentrations are computed by inverting the fitted curve.
- Weighting (1/y, 1/y², etc.) can help handle variance across concentration ranges.
- Back-calculated standards and residuals help reveal bias before you report unknown concentrations.
Values are guides. Also check outliers, standard ranges, dilution factors, and whether each unknown is inside the calibrated range.
FAQ
Should I use 4PL or 5PL?
4PL is often sufficient. Use 5PL if asymmetry is needed (compare with AIC, etc.).
Do I need weighting (1/y, 1/y²)?
Not required. Start without weighting and try it if bias appears across the range.
How should I handle concentration 0 (blank)?
By default, blanks are excluded from fitting (works well with log x). Use blank subtraction only if needed.
An unknown value is outside the standard range.
It will be shown as extrapolated. Accuracy drops, so dilute and remeasure within range.
How do I know whether an unknown concentration is reportable?
Prefer values inside the standard range with acceptable recovery and no extrapolation flag. If an unknown is outside the calibrated range, dilute or rerun the sample so the signal falls within the standards.
Should I change weighting before changing the model?
Usually yes. First compare no weighting with a reasonable weighting option, then review residuals. Move from 4PL to 5PL only when asymmetry remains and the extra parameter improves interpretation.
Interpretation and next steps
Keep this page focused on the ELISA fit itself. Move to the guide when the question is whether blank subtraction belongs in the workflow, whether 5PL really improves the interpretation, or how to explain extrapolated unknowns in a report.
- Assay standard curves guideRead this next for blank handling, 4PL vs 5PL choice, extrapolation checks, and report-ready wording across assay workflows.
- Protein standard curve calculator (BCA / Bradford)Switch here when the workflow is a protein assay standard curve rather than an ELISA plate with logistic fitting.
- A260 concentration & purity calculatorUse this before the assay when sample concentration or purity still needs a quick check.
- Linear regression calculatorOpen a simpler regression page when you want to inspect fit and residual behavior outside the full ELISA workflow.
Comments
Leave questions, requests, or corrections in the comments.