Torsten 0.84

Torsten is a library of C++ functions that support applications of Stan in Pharmacometrics. The current version provides:

• Analytical solution for the one and two linear compartment model with a first-order absorption
• Matrix exponential solution for a general linear compartment model
• Numerical solution for general compartment model using built-in integrators:
  • Runge-Kutta 4th/5th (rk45) order for non-stiff ODE systems
  • Backward Differentiation (bdf) for stiff ODE systems
• Mixed solver for general compartment model with a forcing PK function:
  • Computes analytical solutions for forcing PK One and Two compartment model with a first-order absorption
  • Computes numerical solution for the forced compartment model using the rk45 or bdf method

This prototype is still under development and has been uploaded to facilitate working with the community of Stan developers. The current version was written by Charles Margossian (@charlesm93), Yi Zhang (@yizhang-cae), Bill Gillespie (@billgillespie), and Metrum Research Group, LLC (@metrumresearchgroup. We have recieved extensive help from the Stan development team.

See the user manual (torstenManual.pdf) for more information and guidance on the examples. If you have any questions, please raise an issue on GitHub or send us an e-mail at billg@metrumrg.com.

Licensing

The Torsten library is open-source and licensed under the BSD 3-clause license.

Install

To install CmdStan with Torsten, run the shell script script setupTorsten.sh. To install RStan with Torsten, install the R package TorstenHeaders (https://github.com/metrumresearchgroup/TorstenHeaders) and run install_torsten():

devtools::install_github('metrumresearchgroup/TorstenHeaders')
library(torstenHeaders)
install_torsten()

We are working with Stan's development team to create a system to add and share Stan packages. In the mean time, users can download a forked version of Stan with Torsten from GitHub. The latest version of Torsten (v0.84) is compatible with Stan v2.17.1. Torsten is agnostic to which Stan interface you use.

Examples

For each model, we provide the following files:

• modelName.stan
• modelName.data.R
• modelName.init.R
• modelNameSimulation.R

The simulation file can be used to create the data and initial estimate files.

There are three Stan files for the two compartment model: TwoCptModel, LinTwoCptModel, and GenTwoCptModel. We also provide more sophisticated PK/PD examples. TwoCptModelPopulation extends the two compartment model to the case where we have data from multiple patients with inter-individual variability.

See the manual for more assistance.

Under the R directory, we provide scripts to run the Stan models.

C++ Code

The C++ code for Torsten can be found on the following repos:

Math Library: https://github.com/metrumresearchgroup/math

Stan Grammar: https://github.com/metrumresearchgroup/stan