Woojin Jung, Quyen Thi Tran, Thi Lien Ngo, Jung-woo Chae, Hwi-yeol Yun
College of Pharmacy, Chungnam National University, Daejeon, South Korea
Objectives: Nonlinear mixed effect modeling (NONMEM) for pharmacologic problem is a multidiscipline, highly computer-power demanding and time-spending task, so that the number of pharmacometrics groups carry out NONMEM tasks utilizing their server to overcome mentioned limitations. Nevertherless, there has been not only a struggling point in distributing computer resources between researchers, but it is also hard to maintain the software suitability like version controls among pharmacometricians. As these request from field, the software Docker could be possible solution for server resource management and system suitability as well. It generates a Linux-based virtual environment for each container, enables easy computer resource managing. In addition, the containerized approach has an advantage on automating the whole installation process for pharmacometrics related software, this can reduce unexpected efforts on harmonizing the settings between researchers. In this study, a docker-based server management solution is suggested using programming language R and NONMEM software with its workbench included in integrated development environment (IDE). The workbench aims to provide good compatibility under IDE with remote accessibility and provide essential diagnostics for a quick decision-making supports in model development. Most of the elements in the solution are made with open-source software, for its sustainable development.
Methods: NONMEM (version 7.5.0.) and PsN (Perl-speaks-NONMEM, version 5.2.6.) [1-2] is chosen for performing the modeling. R (version 4.1.2) and Rstudio server as IDE. Docker (version 4.2.0.) image contains R in Rstudio server then NONMEM is added into the environment to be loaded from Rstudio. For management of NONMEM runs and data diagnostics, an R shiny application was made as a package and included as Rstudio add-in.
Results: NONMEM and PsN is fully functional in docker’s virtual environment, compatible with Rstudio server. A package was made to manage NONMEM via PsN commands. It provides model summaries on working directory and xpose’s (R package, version 0.4.13)[3-5] visual diagnostic methods were integrated in graphical user interface (GUI). GUI was built with the framework of R shiny. Basic goodness of fits (GOF), individual plots and visual predictive check (VPC), estimation summary and data inspector were provided in ‘compartmental analysis’ tab. Plus, NCA function from ncappc is added in ‘non-compartmental analysis’ tab. Extensions of “.cov”, “.cor”, “.phi”, “.ext” which is generated from NONMEM run were made to be visualized. Multiple runs from NONMEM can be manipulated in ‘terminal manager’ tab. The diagnostics visualized were positioned in consideration of model developmental flow, choosing scrolling-down method. HTML report generation provides an intuitive/detailed data inspection by its interactivity without any special preparation. Web service is successfully working with the configuration above.
Conclusions: A docker container for NONMEM and its assist program is made. The container can deal with multiple NONMEM runs with graphical managements and container can be remotely accessed. Server can allocate adequate resources by giving the container designated CPUs and memories by docker. Further refinements on generating report and interactive functions are going to be made. Github – https://github.com/tnzo12/rnw
Acknowledgement: This research was funded by Chungnam National Universiy and an Institute of Information and Communications Technology Planning and Evaluation grant funded by the government of Republic of Korea (MSIT; No. 2020-0-01441, Artificial Intelligence Convergence Research Center, Chungnam National University) and supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT)(No.2022R1A2C1010929)
References:
[1] Lindbom L, Pihlgren P, Jonsson EN. PsN-Toolkit–a collection of computer intensive statistical methods for non-linear mixed effect modeling using NONMEM. Comput Methods Programs Biomed. 2005 Sep;79(3):241-57.
[2] Lindbom L, Ribbing J, Jonsson EN. Perl-speaks-NONMEM (PsN)–a Perl module for NONMEM related programming. Comput Methods Programs Biomed. 2004 Aug;75(2):85-94.
[3] Keizer RJ, Karlsson MO, Hooker AC. Modeling and Simulation Workbench for NONMEM: Tutorial on Pirana, PsN, and Xpose. CPT Pharmacometrics Syst Pharmacol 2013, 2: e50.
[4] Jonsson EN, Karlsson MO. Xpose–an S-PLUS based population pharmacokinetic/pharmacodynamic model building aid for NONMEM. Comput Methods Programs Biomed 1999, 58(1): 51-64.
[5] Andrew C. Hooker, Mats O. Karlsson, Justin J. Wilkins and E. Niclas Jonsson (2014). xpose4: Tools for Nonlinear Mixed-Effect Model Building and Diagnostics. R package version 4.X.X. http://xpose.sourceforge.net.
Reference: PAGE 30 (2022) Abstr 9983 [www.page-meeting.org/?abstract=9983]
Poster: Methodology - Other topics