1. Define the reference-to-test comparison
Name the reference and test preparations, assigned reference potency, concentration units, response direction, endpoint, preparation factors, and intended reported unit. Then identify the approved comparison model and uncertainty convention.
Those choices define the result before any curve is fitted. A midpoint ratio, jointly fitted nonlinear estimate, and parallel-line estimate are not interchangeable calculations.
2. Map the wells and fit both preparations
Preserve every raw response with plate position, sample role, replicate, dilution, calculated concentration, transformation, and inclusion state. Confirm the dilution direction and units before fitting.
Apply the method-defined equation, restrictions, weighting, bounds, and convergence rule. Retain observed and fitted values, residuals, parameters, uncertainty inputs, warnings, and every exclusion with its reason.
3. Decide whether one potency comparison is meaningful
Evaluate the reference and test relationship using the predefined diagnostic appropriate to the model: for example, linearity and parallelism for a parallel-line method or independently fitted shape evidence for a common-shape logistic comparison.
A joint calculation can encode common shape by design; it cannot prove that assumption with the same shared parameters. Visual similarity helps a reviewer understand the curves, but the approved independent diagnostic supplies the decision.
4. Calculate potency and uncertainty
Calculate with stored, unrounded values using the approved direction, reference assignment, preparation corrections, and model convention. Report the estimate and unit beside its confidence interval or other method-defined uncertainty.
The worked guide carries the increasing common-shape 4PL example from reference EC50 20.000 ng/mL and test EC50 25.000 ng/mL to 80.0%. This hub keeps the broader workflow separate from that arithmetic lesson.
5. Apply every predefined suitability rule
Evaluate reference performance, data and replicate checks, model diagnostics, comparability, result range, and other required criteria in their controlled version. Overall suitability passes only under the method’s stated combination rule.
A failed or not-evaluable required criterion remains visible and produces the configured overall outcome. The quality procedure determines investigation, repeat, invalidation, or other disposition; the calculation does not rewrite the failure.
6. Report the result and the evidence that produced it
Connect the final estimate and uncertainty to source identity, well mapping, concentrations, transformations, model version, restrictions, diagnostics, exclusions, suitability rules and observed values, software version, and source and configuration fingerprints.
Show the number, comparison, limit, unit, and PASS, FAIL, or not-evaluable status for every criterion. A colored label alone is not a reviewable result.
Limits and adjacent decisions
Model selection belongs in method development; the linked comparison guide owns that decision. Criterion selection and thresholds belong in development and validation; the linked suitability guide carries one rule set through overall disposition.
Provenarium supports three exact supplier-verified quantitative relative-potency contracts: parallel-line log-log, common-shape 4PL, and common-shape 5PL. It does not provide routine model search or silently substitute one contract when another fails.
Frequently asked questions
Does relative potency always equal the ratio of EC50 values?
No. That ratio is a useful illustration under compatible curve and concentration conventions. Parallel-line, slope-ratio, jointly fitted nonlinear, or preparation-adjusted methods may define the estimate differently.
What happens when a suitability criterion fails?
The configured outcome should be displayed and reported as a failure. The organization’s procedure determines whether the run is invalidated, investigated, repeated, or otherwise handled; software should not invent that decision.