Common-shape 5PL relative potency

What this model estimates

The preparations share lower and upper asymptotes, slope and asymmetry, with preparation-specific horizontal locations. The joint horizontal shifts give test-to-reference relative activity and absolute potency. The model permits an asymmetric shape on the fitted axes; it does not give every preparation an unrelated shape.

Use 5PL when your justified assay model and data support estimating that extra shape parameter. More parameters do not by themselves provide better or more reliable potency.

Data and design needed

Provide one reference, at least one test, at least six common positive dose levels and an equal count of at least two consecutively numbered replicates in every preparation–dose cell after exclusions.

Include informative curve coverage rather than six tightly clustered concentrations. The fitted shape, potency and asymmetry must be identifiable, with finite positive within-cell pure-error variance. Six levels are a minimum, not a guarantee that asymmetry can be measured.

Check the settings

1. Inspect the map, dose series, reference and units. Select Common-shape 5PL.
2. Choose Untransformed positive doses or one of the three logarithmic dose transformations. Response may also be untransformed or logarithmic. Review the justified Weighting and its response-domain restrictions.
3. Review saved suitability limits and the intended result before Run analysis. The model’s fitting and asymmetry checks are fixed and cannot be adjusted in the editor.

Read potency and the asymmetry assessment

Read each test’s potency and two-sided 95% normal linearized Fieller interval in the PDF report, together with the model diagnostics and fixed adequacy outcomes. The interval uses replicated-cell pure error and joint horizontal-shift covariance for every weighting option.

The 95% asymmetry profile must be finite, connected and inside the computational bounds. A profile that includes asymmetry 1 is reported as asymmetry not demonstrated; exclusion of 1 indicates detected asymmetry. This is informational and is not automatic selection between 4PL and 5PL.

Failures and limitations

A flat, unbounded or otherwise invalid asymmetry profile produces a weak-identifiability failure. A result may also be unavailable if the design is insufficient, transformations or weights are invalid, the fit does not converge, pure error is zero or Fieller limits cannot be calculated.

Do not interpret a failed asymmetry assessment as an approved 4PL result. Correct a demonstrable input error or investigate the design under your procedure; a new model choice requires justified new development work.

• Scientific background: Gottschalk and Dunn (2005), 5PL characterization