ShinyMixR: A project-centric R/Shiny run management tool for nlmixr
Richard Hooijmaijers (1,5), Matt Fidler (3,5), Rik Schoemaker (2,5), Mirjam N. Trame (3,5), Wenping Wang (3,5), Justin J. Wilkins (2,5), Yuan Xiong (4,5), Teun M. Post (1,5)
(1) LAP&P Consultants, The Netherlands, (2) Occams, The Netherlands, (3) Novartis Pharmaceuticals, USA, (4) Certara Strategic Consulting, USA, (5) The nlmixr team
Objectives: The combination of open-source packages nlmixr and RxODE, available on CRAN [1, 2] and actively developed on GitHub (http://github.com/nlmixrdevelopment/nlmixr) [1, 2], provides a non-linear mixed effects system to perform population-type pharmacokinetic and pharmacodynamic analyses and simulations [3] in R [4]. The ability to perform population modeling in R provides an opportunity to work via a single unified workflow for data management, data exploration, data analysis and report writing. The aim of this current work was to develop a user-friendly tool for nlmixr based on Shiny, which would facilitate a workflow around an nlmixr project. Ultimately, this should allow for: 1) dynamic and interactive model development, 2) quick and efficient communication of population PK-PD models, 3) rapid demonstration of simulation results from PK and PK-PD modelling (also see RxODE Shiny), and 4) reporting of modelling results [5].
Methods: ShinyMixR [6]is set up as an open source nlmixr project management tool written completely in R, and deployed as a set of R functions. The ShinyMixR system is built around a project-centric structure and provides an interface to nlmixr from both the R command line (R, related GUIs and RStudio [7]) as well as a user-friendly Shiny dashboard application [8]. The ‘shinydashboard’ package [9] provides a layer on top of Shiny to produce an easy-to-use dashboard which can be used for controlling and tracking runs with an nlmixr project, and was the basis for setting up the modular interface. Most of the functions underlying the interface are written such that these can be called independently from the R command line, and also work in combination with the graphical interface and vice versa.
Results: Using the ShinyMixR package, the user can specify and control an nlmixr project workflow entirely in R. Within a project folder, a structure can be created to include separate folders for models, data and runs. Functionality is available to edit and run model code, summarize and compare model outputs in a tabular fashion, and view model development using a tree paradigm. Inputs, outputs and metadata are stored in relation to the model code within the project structure (a discrete R object) to ensure traceability. The results can be visualized by using modifications of existing packages (such as xpose.nlmixr [10]), user-written functions and packages, or pre-existing plotting functionality included in the ShinyMixR package. Results can be reported using the R3port package in pdf and html format [5]. In this project-oriented structure, the command line and dashboard can be used independently and/or interdependently. As such, projects are set up to be user-centric, instead of interface-dependent.
Conclusions: The ShinyMixR package provides a means to build a project-centric workflow around nlmixr from the R command line and from a streamlined Shiny application. This project tool was developed to enhance the usability and attractiveness of nlmixr, facilitating dynamic and interactive use in real-time for rapid model development.
References:
[1] CRAN: https://cran.r-project.org/web/packages/nlmixr/index.html and GitHub: https://github.com/nlmixrdevelopment/nlmixr
[2] CRAN: https://cran.r-project.org/web/packages/RxODE/index.html and GitHub: https://github.com/nlmixrdevelopment/RxODE
[3] Wang W et al. CPT:PSP (2016) 5, 3–10.
[4] R Core Team (2015). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/
[5] https://github.com/RichardHooijmaijers/R3port
[6] https://github.com/RichardHooijmaijers/shinyMixR
[7] RStudio Team (2015). RStudio: Integrated Development for R. RStudio, Inc., Boston, MA URL http://www.rstudio.com/
[8] http://shiny.rstudio.com/
[9] https://cran.r-project.org/web/packages/shinydashboard/shinydashboard.pdf
[10] https://github.com/nlmixrdevelopment/xpose.nlmixr
