Phoenix WinNonlin Validation Tests

Phoenix WinNonlin validation tests

The Phoenix WinNonlin Validation Suite provides automated tests to validate Phoenix WinNonlin functions and options, including PK, PD, PKPD, and Indirect modeling, and data manipulation. Following an automated test run, the Phoenix WinNonlin Validation Suite generates a report that details the tests and results.

For a complete list of tests, see “WNL Test cases”.

The three base tests include an installation test case to verify that all Phoenix files have been installed in the correct locations and that they have not been modified. The other two base test cases verify that 1) two CSV files known to match will correctly be passed when compared on the user’s system and 2) two CSV files known to differ will correctly be failed when compared on the user’s system. These two test cases verify that the Validation Suite tool is functioning as intended.

Note:The Phoenix WinNonlin Validation Suite includes test cases for the Phoenix Model Object with individual (non-population) modeling. The Phoenix Model Object (NLME engine) can only run on 64-bit systems, so any test cases for the Phoenix Model Object will have a status of “not applicable” (N/A) on 32-bit systems. These tests cases are: PHX LM 501, PHX MM 301, PHX PD 103, PHX PD 105, PHX PK 1 Clearance, PHX PK 1 PD 101, PHX PK 12, PHX PK 18, PHX PK 6, PHX PK 9.

WNL Test cases

Framework Tests

PHX Installation: Proper installation of all files in the correct location.

CSV File Comparison Success: Correct identification of identical CSV files by the comparison tool.

CSV File Comparison Failure: Correct identification of differing CSV files by the comparison tool.

WinNonlin Tests

PHX Desc Stats No Weighting: Descriptive Statistics with no weighting, including confidence intervals, specified number of SD, standard percentiles, and user-defined percentiles.

PHX Desc Stats Weighting: Descriptive Statistics with weighting, including confidence intervals and units.

PHX DW Transformation Arithmetic: Data Wizard Transformation with each of the predefined arithmetic transformations.

PHX DW Transformation Baseline: Data Wizard Baseline Transformations including change from fixed value, change from minimum per group, percent change from fixed value, and ratios using minimum per group.

PHX DW Transformation Functions: Data Wizard Transformation to perform inverse, absolute, exponential, ln, log10, and square root calculations.

PHX Indiv LM 501: Non-population Phoenix Model with constant linear model. This test case is not applicable when run on a 32-bit system.

PHX Indiv MM 301: Non-population Phoenix Model with one compartment, bolus input, and Michaelis-Menten output. This test case is not applicable when run on a 32-bit system.

PHX Indiv PD 103: Non-population Phoenix Model with an inhibitory effect E0 model. This test case is not applicable when run on a 32-bit system.

PHX Indiv PD 105: Non-population Phoenix Model with Sigmoid Emax model. This test case is not applicable when run on a 32-bit system.

PHX Indiv PK 1 Clearance: Non-population Phoenix Model with one compartment, IV bolus input, and first-order output with clearance parameterization. This test case is not applicable when run on a 32-bit system.

PHX Indiv PK 1 PD 101: Non-population Phoenix Model with an effect compartment PK/PD link model. This test case is not applicable when run on a 32-bit system.

PHX Indiv PK 12: Non-population Phoenix Model with two compartments, first-order input, first-order output, lag time, and with micro-constants as primary parameters. This test case is not applicable when run on a 32-bit system.

PHX Indiv PK 18: Non-population Phoenix Model with three compartments, IV bolus input, first-order output, and with macro-constants as primary parameters. This test case is not applicable when run on a 32-bit system.

PHX Indiv PK 6: Non-population Phoenix Model with one compartment, equal first-order input and output, and with lag time. This test case is not applicable when run on a 32-bit system.

PHX Indiv PK 9 Units: Non-population Phoenix Model with two compartments, constant IV infusion input, first-order output, and with micro-constants as primary parameters, with units for the primary and secondary parameter results. This test case is not applicable when run on a 32-bit system.

PHX Ratios and Differences: Ratios and Differences from one input worksheet, from two input worksheets, and using means of non-unique values.

WNL Bioeq Average Crossover 2x2: Average Bioequivalence using the default models for non-replicated two-period crossover data, both the fixed and random effects default model and the all fixed-effects default model.The fixed-effects model also specifies numerator and denominator for an additional F-test.

WNL Bioeq Average Crossover 2x4: Average Bioequivalence using the default model for replicated crossover data, in this case with four periods.

WNL Bioeq Average Parallel: Average Bioequivalence using the default model for parallel data.

WNL Bioeq Population Individual: Population and individual Bioequivalence using the predefined model.

WNL Crossover Separate: Crossover nonparametric analysis using separate input data columns.

WNL Crossover Stacked: Crossover nonparametric analysis using stacked input data columns.

WNL Deconv 1exp None TimesFromCol Units: Deconvolution using one-exponent unit impulse response (UIR) function, with no smoothing, output times from a worksheet, and units.

WNL Deconv 2exp User Rate Change ToLastPt: Deconvolution using two-exponent UIR function with specified smoothing parameter, initial rate and initial change in rate set to 0, and output times generated from 0 to the last observed time point.

WNL Deconv 3exp Auto Rate NumPts FromTo: Deconvolution using three-exponent UIR function, using the Phoenix-determined optimal smoothing parameter value, initial change in rate set to 0, and output times generated for specified time range.

WNL LinMix Ancova Options: Linear Mixed Effects using ANCOVA model (classification and regressors), with an interaction term, random intercept, dependent variable transformation, contrasts, and estimates.

WNL LinMix AR ARH CSH FA0: Linear Mixed Effects models with autoregressive, heterogeneous autoregressive, heterogeneous compound symmetry, and no-diagonal factor analytic types of covariance structure, one model tests no intercept for fixed effects.

WNL LinMix CS TOEP UN: Linear Mixed Effects models with compound symmetry, Toeplitz, and unstructured types of covariance structure, and with using a weight variable.

WNL LinMix VC Contrasts Estimates LSMs: Linear Mixed Effects models with variance components covariance structure type, contrasts, estimates, least squares means and pairwise differences, and residual DF.

WNL LM 501 Constant: Least-Squares Regression Modeling, Linear model 501 (constant).

WNL LM 502 Linear: Least-Squares Regression Modeling, Linear model 502 (linear).

WNL LM 503 Quadratic: Least-Squares Regression Modeling, Linear model 503 (quadratic).

WNL LM 504 Cubic: Least-Squares Regression Modeling, Linear model 504 (cubic).

WNL MM 301: Least-Squares Regression Modeling, one compartment model with IV bolus input and Michaelis-Menten output.

WNL MM 302: Least-Squares Regression Modeling, one compartment model with constant IV input and Michaelis-Menten output.

WNL MM 303: Least-Squares Regression Modeling, one compartment model with first-order input and Michaelis-Menten output.

WNL MM 304: Least-Squares Regression Modeling, one compartment model with first-order input, Michaelis-Menten output, and a time lag.

WNL NCA Extravascular NonSS PartialAUC: NonCompartmental Analysis for non-steady state, plasma, extravascular data, using the linear trapezoidal linear interpolation calculation method and including partial areas.

WNL NCA Infusion NonSS PartialAUC Rules: NonCompartmental Analysis for non-steady state, plasma, IV infusion data, using the linear trapezoidal linear interpolation calculation method and Lambda Z Best Fit rules, and including partial areas.

WNL NCA IV Bolus NonSS PartialAUC LinearLog Trap: NonCompartmental Analysis for non-steady state, plasma, IV bolus data using the linear log trapezoidal calculation method and including a partial area.

WNL NCA IV Bolus SS InsertC0 UserDefined: NonCompartmental Analysis for steady state, plasma, IV bolus data, using the linear trapezoidal linear interpolation calculation method and including methods for inserting C0, partial areas, user-defined parameters, and concentrations (computed at specified times).

WNL NCA PartialAUC LinearUpLogDown Units: NonCompartmental Analysis for plasma, extravascular data, using the linear up log down calculation method and including partial areas and units.

WNL NCA PartialAUC LinTrapLinLogInterp Units: NonCompartmental Analysis for plasma, extravascular data, using the linear trapezoidal linear/log interpolation calculation method and including partial areas and units.

WNL NCA PD UserDefined: NonCompartmental Analysis for drug effect data, using the linear trapezoidal linear interpolation calculation method, with baseline and threshold values, user-defined time ranges for slopes, partial area, user-defined parameter, and effects computed at specified x-values.

WNL NCA SparseSampling Extravascular: NonCompartmental Analysis for sparse-sampling, plasma, extravascular data, using the linear trapezoidal linear interpolation calculation method, includes case with missing values.

WNL NCA Urine TimeRange Flags Sort: NonCompartmental Analysis for urine data, using the linear trapezoidal linear interpolation calculation method, with Lambda Z acceptance criteria, user-defined time ranges for slopes, and a sort variable.

WNL NCA Weighting InsertC0: NonCompartmental Analysis for plasma, IV bolus data, using the linear trapezoidal linear interpolation calculation method, and including methods for inserting C0, partial areas, 1/Y weighting, and a sort variable.

WNL NPS Regular Linear BestFit: NonParametric Superposition with dosing at regular intervals and using linear interpolation.

WNL NPS Variable Log StartEnd: NonParametric Superposition with variable dosing intervals, a specified output time range, and using log interpolation.

WNL PD 101: Least-Squares Regression Modeling, simple Emax model.

WNL PD 102: Least-Squares Regression Modeling, simple Emax model with baseline effect.

WNL PD 103: Least-Squares Regression Modeling, inhibitory effect Emax model.

WNL PD 104: Least-Squares Regression Modeling, inhibitory effect model with a baseline effect parameter.

WNL PD 105: Least-Squares Regression Modeling, sigmoid effect model without a baseline effect parameter.

WNL PD 106: Least-Squares Regression Modeling, sigmoid Emax model with baseline effect.

WNL PD 107: Least-Squares Regression Modeling, sigmoid inhibitory effect model.

WNL PD 108: Least-Squares Regression Modeling, sigmoid inhibitory effect model with a baseline effect parameter.

WNL PK 1 Clearance: Least-Squares Regression Modeling, one compartment PK model with IV bolus input and first-order output with clearance parameterization.

WNL PK 1 PD 101 Linked: Least-Squares Regression Modeling, PK/PD linked model with effect compartment.

WNL PK 10: Least-Squares Regression Modeling, two compartment PK model with constant IV infusion input and first-order output, with macro-constants as primary parameters.

WNL PK 11: Least-Squares Regression Modeling, two compartment PK model with first-order input, first-order output, without lag time, with micro-constants as primary parameters.

WNL PK 12: Least-Squares Regression Modeling, two compartment PK model with first-order input, first-order output, with lag time, with micro-constants as primary parameters.

WNL PK 13: Least-Squares Regression Modeling, two compartment PK model with first-order input, first-order output, without lag time, with macro-constants as primary parameters.

WNL PK 14: Least-Squares Regression Modeling, two compartment PK model with first-order input, first-order output, with lag time, with macro-constants as primary parameters.

WNL PK 15: Least-Squares Regression Modeling, one compartment PK model with simultaneous IV bolus and constant IV infusion.

WNL PK 16: Least-Squares Regression Modeling, two compartment PK model with simultaneous IV bolus and constant IV infusion input, with micro-constants as primary parameters.

WNL PK 17: Least-Squares Regression Modeling, two compartment PK model with simultaneous IV bolus and constant IV infusion input, with macro-constants as primary parameters.

WNL PK 18: Least-Squares Regression Modeling, three compartment PK model with IV bolus input, first-order output, with macro-constants as primary parameters.

WNL PK 19: Least-Squares Regression Modeling, three compartment PK model with constant IV infusion, with macro-constants as primary parameters.

WNL PK 2: Least-Squares Regression Modeling, one compartment PK model with constant IV infusion input and first-order absorption.

WNL PK 3 Clearance: Least-Squares Regression Modeling, one compartment PK model with first-order input and output, without lag time, but with clearance parameterization.

WNL PK 4 Clearance: Least-Squares Regression Modeling, one compartment PK model with first-order input and output, with lag time and clearance parameterization.

WNL PK 5: Least-Squares Regression Modeling, one compartment PK model with equal first-order input and output, without lag time.

WNL PK 6: Least-Squares Regression Modeling, one compartment PK model with equal first-order input and output, with lag time.

WNL PK 7 Clearance: Least-Squares Regression Modeling, two compartment PK model with IV bolus input and first-order output, with micro-constants as primary parameters and with clearance parameterization.

WNL PK 8: Least-Squares Regression Modeling, two compartment PK model with IV bolus input and first-order output, with macro constants as primary parameters.

WNL PK 9: Least-Squares Regression Modeling, two compartment PK model with constant IV infusion input and first-order output, with micro-constants as primary parameters.

WNL SCM Linear: Semicompartmental Modeling using a linear piecewise PK model.

WNL SCM LinearLog: Semicompartmental Modeling using a PK model that is linear to Tmax and then log-linear after Tmax.

WNL SCM Log: Semicompartmental Modeling using a log-linear piecewise PK model.