Set up the reference and test series explicitly
Assume a reference standard and test sample are each measured at eight dilution levels with three replicate wells. Record the initial concentration, dilution factor, concentration unit, sample role, replicate, and well. Do not infer sample identity from color or column position after analysis.
The reference may carry an assigned potency or activity. Preserve that value and its unit separately from the nominal concentration used on the plate. Preparation and predilution factors must be traceable into the final calculation.
| Input | Reference example | Test example |
|---|---|---|
| Role | Reference standard | Test sample |
| Fitted midpoint | 20.000 ng/mL | 25.000 ng/mL |
| Assigned factor | 100% | Not applicable |
| Curve convention | Increasing 4PL | Increasing 4PL |
Fit curves under the method’s comparison model
An illustrative workflow fits a 4PL to each sample and reviews whether curve behavior supports comparison. A production method may instead use a joint nonlinear model with shared parameters, a restricted fit, a parallel-line model, or another validated procedure.
Review reference performance first. Then evaluate test curve coverage, residuals, replicate variability, and any parallelism or equivalence rule defined in the method. An attractive ratio does not rescue an unsuitable comparison.
Calculate and label the ratio
Under the increasing-curve convention used here, the test requires a higher concentration than the reference to reach its midpoint, so its relative potency is lower. Dividing 20.000 by 25.000 and multiplying by 100 gives 80.0%, matching the result in the downloadable worked-example report.
If the assay curve decreases, the software equation reverses parameter direction, or the standard carries an assigned potency other than 100%, verify the method-specific formula. Apply preparation corrections at the defined stage and retain unrounded values for downstream calculations.
20.000 ÷ 25.000 × 100 = 80.0%Calculate the midpoint ratio
This guide uses reference EC50 ÷ test EC50 × assigned reference potency. Change the values to see how each fitted midpoint affects the result.
- Illustrative relative potency
- 80.0%
A reportable potency result also requires reference and test curves that support comparison and the system-suitability checks defined for the assay.
Follow one 80.0% result from plate data to report.
Use the source CSV, reusable analysis configuration, and finished PDF to check how the reference and test curves produce the reported result. You can also open the same plate in Provenarium and repeat the analysis yourself.
Report suitability next to the result
Each criterion should show its name, definition, observed value, units where relevant, configured limit, and PASS or FAIL. Conditional formatting can make the outcome scannable, but color must not be the only carrier of meaning.
Common categories include reference curve performance, response range, parameter bounds, residual or lack-of-fit evidence, replicate variability, model comparison, and relative-potency range. Select and justify criteria for the assay; the example does not establish universal thresholds.
Make the worked result reproducible
The report package should include analyst and date, source data file fingerprint, selected spreadsheet sheet, plate map, dilution table, long-form observations, exclusions and reasons, model and parameter table, fitted plot, suitability table, reported result, software version, and analysis configuration fingerprint.
Export the reusable analysis configuration separately from the source data. Applying that configuration to another compatible plate preserves the worksheet selection, plate assignments, dilution scheme, model, suitability criteria, and report settings without carrying over the original measurements or source-file identity.
Frequently asked questions
Which EC50 belongs in the numerator?
That depends on the model and reporting convention. In this increasing-curve example, reference EC50 divided by test EC50 yields test potency relative to reference. Encode and label the convention rather than relying on memory.
Can I calculate relative potency when curves are not parallel?
A ratio can always be computed numerically, but it may not be scientifically meaningful under a model requiring comparable shapes. Follow the method’s comparability assessment and failure handling.
Should rounded EC50 values be used in the ratio?
No. Calculate with full stored precision and round only the displayed result according to the reporting rule.