List of known bugs in MonolixSuite 2024R1

The list of known bugs in every version of MonolixSuite can be found in the release notes of the next version.

We list below the known bugs in MonolixSuite 2024R1 (they will be fixed in the next version). The 2024R1 version of Monolix can be downloaded here.

All applications

• [Data] When a column is tagged as SS and a column is tagged as ADDL, the setting “number of steady-state doses” does not appear. The workaround is to first ignore the ADDl column, set the number of doses and then tag the ADDL column.

• [Monolix & PKanalix / Plots] In the Individual fits plot, editing the stratification groups for a continuous covariate has the effect of hiding the pages for subplots such that it is no longer possible to move to another page until pagination is reset or changed by the user.

• [Monolix & PKanalix / Reporting] Plot content outside of custom limits are displayed in the report over the axes.

• [Plots] In a plot stratified by groups with the “merged splits” option, the coloring by group is not reset after unselecting “merged splits”.

• [Documentation] In the Documentation tab in Home, reading videos does not work.

• [Data] Loading a dataset or a model file with a single quote (') in the filename should not be accepted, and leads to an error at reload: "Parsing failed at line x, column x, due to incomplete match."

• [Data] Loading an excel file located in a path containing japanese characters leads to the error “Impossible to load XXX.xlsx”.
  Workaround: save the excel file as csv and load the csv file instead

• [Reporting] Plots not exported correctly when the “start” argument of “mainLayout” is not a multiple of “rows” multiplied by “cols”.

Monolix

• When changing the structural model for a model with additional observation ids, the mapping for the additional observation ids is not done automatically.

• [Bootstrap] After running bootstrap on a project where there are effects of several covariates on the same parameter, with at least one of the covariates that is transformed, the beta values for covariate effects are swapped in the table of estimated population parameters in Estimation compared to the bootstrap results.

• [Bootstrap] If there are additional unused observation ids in the dataset which are not present for all individuals, bootstrap may stop with the error: Observation id () is missing from the resampled dataset and cannot be mapped to the model output. Workaround: filter out unused observation ids from the dataset.

• [Bootstrap] Just after running bootstrap or loading a project with bootstrap results, clicking on the legend of the Pop Median plot does not work. Workaround: it works after going to another plot and back.

• [TGI library] The TGI model named TG_Exp_NoFeat_TS0par_TGI_t0Trt_LK_Const_Claret_SD_NoFeat includes a parameter tau that is not used in the model, while it should be used to define a delay. (This bug was already there in the 2023 version)

• Typical values of population parameters by categories of categorical covariates are not displayed for latent covariates.

• copyData is an option available in config files to run model building, but not for convergence assessment even though the option is available in the GUI.

• [Obs versus Pred plot] When all observations and predictions are negatives (this can happen when working with observations which are already log-transformed), the identity line is not shown.

• [Obs versus Pred plot] Incorrect x limits are sometimes used for the identity line in log scale.

• [Individual fits plot] Sorting individual fits by individual parameters can be wrong in case of parameters without IIV, as the index of selection applies to the list including parameters without IIV.

• Exported plots in ChartsFigures are deleted when running the Plots task again or when loading the project with enabled "Create plots at project load" Preference.

• When saving a Monolix project under a new name with existing plots saved in the ChartsFigures subfolder of the results folder, these figures are copied in the new results folder even if the project has been modified and the results are now invalid.

• [reporting] when a placeholder has no colors specified, it takes colors from the pevious placeholder instead of the default colors from the project

• [BLQ predictive check] The plot can appear cut off in the GUI, with no option to adjust the axes manually. To resolve this, close and reopen the project, then regenerate the plots.

• [Export to Simulx] Exporting a Monolix project with Filters leads to a crash of the Monolix application.
  Workaround: Instead of doing an “Export to” in Monolix, do an “Import from” in Simulx.

• [Export to Simulx] Exporting a Monolix project having (1) data formatting and (2) a non-continuous type of data (event, count or categorical) leads to an error "Impossible to create a valid data / model fit because the data set and the structural model are inconsistent".
  Workaround: Click Export > Export formatted dataset to generate the dataset with formatting options as csv file, and load this new dataset. Rerun or use last estimates and then export to Simulx.

• [Export to Simulx] Exporting a Monolix project having (1) Filters defined based on a covariate and (2) not selecting this covariate in the export menu can lead to an error in Simulx.
  Workaround: export the covariates used to define Filters

PKanalix

• Clicking on Check lambda_z tab triggers a star (unsaved changes) even though the project has not changed.

• Exporting to Simulx a PKanalix project where a covariate is used in a custom NCA parameter, without that covariate, generates an error in Simulx

• Exporting to Simulx a PKanalix project where units are defined using a scaling factor generates an error in Simulx

• Ctau and AUC_TAU are not correctly calculated when there is no observation points at the end of the interval (time=tau) and lambda_z could not be calculated. Instead of being NaN, Ctau is equal to Clast and AUC_TAU is equal to AUC_last. This bug is also present in previous versions. Note that when lambda_z is available, Ctau and AUC_tau are correctly calculated by extrapolating the concentration until time=tau using lambda_z.

• AUC_TAU is not correctly calculated when the three following conditions are met: (1) the option 'inter-dose interval for single dose profiles” is used, (2) there are occasions and thus severeal profiles per individuals, and (3) there is no concentration at the time of the dose. In this case, the AUC_TAU is calculated using C0=0 instead of C0=min concentration over interval.
  Workaround: filter the data in the Filter tab to keep only one profile per individual

• [Plot Individual fits] label for y-axis is not saved when saving the project and yLabel setting in reporting placeholder is ignored

• Sparse NCA gives inconsistent results for custom NCA parameters defined with continuous covariates.

• Results from sparse NCA on urine data are not returned.

• [Reporting] When the dataset has initially several observation types and one has been filtered out with the Filters module, the %datatype% keyword is still replaced by the data types for the full dataset. (This bug was already there in the previous version)

• Loading a dataset with only negative times should work but gives an error: “Value of manual tau must be strictly positive”.

• [Reporting] The accumulation ratio table covariate values are wrong.

• [Reporting] In individual fits, the y-axis label setting is not applied in the placeholder. Even when yLabel: 'Label' is manually set, the argument is ignored.

• [Reporting] Some Word default table styles that are specified directly in the placeholder are not adopted in the preview or report.

Simulx

• [Import] Importing a Monolix or PKanalix project with Filters leading to less occasions than in the unfiltered dataset generates a crash or wrong tables for mlx_XXX elements in Simulx

• [Import] Importing a Monolix or PKanalix project with a single individual and regressors leads the sd of the mlx_RegDist element to be NaN. After having been saved, these simulx projects do not reload.

• [Import] Importing a Monolix project which includes (1) data formatting and (2) non-continuous data leads to a crash.
  Workaround: In Monolix, click Export > Export formatted dataset to generate the dataset with formatting options as a csv file, and load this new dataset. Rerun or use last estimates and then export to Simulx.

• [Import] Importing a Monolix or PKanalix project with a dataset having regressors which are missing (indicated as dots) on some lines and several Obs ID may generate an error "Invalid subject-occasion first regressor value detected at line XXX" when filtering by one ObsId would generate a dataset with only missing regressor values for some individuals.

• [Import] importing a Monolix project with a categorical covariate having a single modality (after a Filter action or in the input dataset) leads to a crash in Simulx upon simulation.
  Workaround: in Monolix, do not tag the covariate with single modality as categorical covariate.

• [Simulation] Running a simulation with different output variables selected for different simulation groups results in a crash.

• [Outcome distribution plot] Switching the outcome distribution display from boxplot to PDF or CDF causes the plot to disappear. Only reloading the project restores the distribution plots as boxplots again.

• [Outcome distribution plot] If an outcome is defined based on a group-specific output element, the simulation group displayed on the outcome plot may be wrong.

• [reload] When reloading a saved Simulx project which was created by importing a Monolix project into Simulx, error “Missing values for the second part”. This happens when in the Monolix project some categorical covariate columns have only a single value, or when the dataset has been filtered such that only one category remains for some categorical covariates.
  Workaround:
  If you don’t use mlx_CovDist, manually delete the line defining mlx_CovDist in the .smlx file. The line number of indicated in the error message.
  If you use mlx_CovDist in the “Simulation” tab, you can add the definition which are missing for the covariates which have a single category.
  If the covariate definition in the model is the following:

  CODE

  <MODEL>
  [COVARIATE]
  input = {TREAT, WTKG, SEX, STDY}
  TREAT = {type=categorical, categories={'A','B'}}
  SEX = {type=categorical, categories='0'}
  STDY = {type=categorical, categories='101'}

  with SEX and STUDY having a single category, the wrong definition of mlx_CovDist will be the following, containing only WTKG and TREAT:

  CODE

  COVARIATE:
  continuousCovariates={WTKG}
  categoricalCovariates={TREAT, SEX, STDY}
  'mlx_CovDist' = {{distribution=logNormal, typical=70, sd=0.3}, {categories={'A', 'B'}, probabilities={0.6, 0.3}}}

  It needs to be extended to include also SEX and STUDY, using the same category as listed in the <MODEL> section and probability 1. Note that the order of the definition must match the order of covariates listed above.

  CODE

  'mlx_CovDist' = {{distribution=logNormal, typical=70, sd=0.3}, {categories={'A', 'B'}, probabilities={0.6, 0.3}}, {categories={'0'}, probabilities={1}}, {categories={'101'}, probabilities={1}}}

Sycomore

• Double-clicking on a Sycomore project opens a new Sycomore session without loading that project.

• The “copy table” action in Comparison tab has several formatting issues with the resulting table pasted in a text file, Excel or Word files. (already there in previous versions)

Connectors

• [Monolix] A VPC split by some groups generated with plotVpc() can show issues in the prediction intervals if the max binLimit is different across groups. The workaround is to use filter instead of split to generate the VPC separately for each group.

• [Monolix] When using plotVpc() with settings linearInterpolation=TRUE and legend=TRUE, or plotResidualsScatterPlot with also predInterval=TRUE, the color for prediction intervals in the legend is red instead of pink.

• [Monolix] When calling setScenario() specifying a partial list of tasks for a project without results, R crashes. Workaroud: call setScenario() with the modified object returned from getScenario() (see doumentation).

• [Simulx] getOutputElements() returns data truncated to 5000 lines, similarly to what is done in the GUI to speed up the display.

• [Simulx] Regressor elements are missing from the output displayed by printSimulationSetup(), although the regressors are correctly applied in the project.

• [PKanalix] Some metrics are missing in the result of getNCAParameterStatistics: Q1, Q3, max, median.

• [Monolix-PKanalix] plotObservedData() with turned on legend, turned off dots and lines and enabled merged split outputs legend with switched modalities.
  Workaround: plot with lines and subsequently remove the lines

  CODE

  p <-plotObservedData(settings = list(legend = T, mean = T, dots = F, lines = T),
                       stratify = list(split = "SEX", mergedSplits = T))
  p$layers[[1]]$aes_params$size = NA
  p

• [Simulx] when exporting a simulx project to Monolix or PKanalix using exportProject() with argument "modelFileName", the modelFilename is not used and the model is saved as model.txt instead.

• [PKanalix] in plotCAObservationsVsPredictions(), when using log-scale and with some observations being zero, the dots with value zero are being shown at the border of the plot instead of being hidden.

• [PKanalix] in plotNCAIndividualFits(), when using splitOccasions=F, the number of subplots is not properly adjusted.

• [Simulx] If a Monolix project has a covariate that is the same for all individuals, then when you initialize connectors for Simulx, and import such a Monolix project, the mlx_cov is not returned with getCovariateElement(). In this case, initialize first connectors for Monolix and export a Monolix project to Simulx.

• [Monolix] In the VPC plot censored data are not colored if the stratification (split) contains only one ID per group.

• [Monolix] In the VPC plot, prediction interval is removed if it is not entirely included in a plot and when "ylim" is used as a setting option. A workaround is to use: coord_cartesian(xlim = c(NA, NA), ylim = c(1, 10000)) as a ggplot post-processing.

• [PKanalix] running getNCAIndividualParameters() just after loadProject() without runNCAEstimation() leads to getNCAIndividualParameters()$information$CDISC containing PKanalix names instead of CDISC names.
  Workaround: call run runNCAEstimation() before running getNCAIndividualParameters()