CRAN status R-CMD-check Codecov test coverage Lifecycle: maturing

Script based ‘NONMEM’ model development in RStudio.


You can install the released version of NMproject from CRAN with:


To install the latest version of NMproject from GitHub:

if(!require("devtools")) install.packages("devtools")

To install a specific release (e.g. v0.5.1) on GitHub use the following command:


Load the package with


Getting started with NMproject

Two options:

  1. Running the demo is easiest way to familiarise your with NMproject.
  2. Reading the website vignette.

Code snippets

Use of pipes, %>%, make it easy to code sequences of operations to model objects.

Following snippet adds covariates to model object, m2:

m2WT <- m2 %>% child(run_id = "m2WT") %>%
  add_cov(param = "CL", cov = "WT", state = "power") %>%

Graphical RStudio ‘Addins’ exist for reviewing the changes that functions like add_cov() make before execution and performing nm_tran() checks.

For more complex operations use fully tracked manual edits.

Apply fully customisable diagnostic reports to one or multiple objects with nm_render() like so:

c(m1, m2) %>% nm_render("Scripts/basic_gof.Rmd")
## Saves html diagnostic reports in "Results" directory

The template Scripts/basic_gof.Rmd can also be run as an R notebook for interactively customising to your specific model evaluation criteria.

Here’s a snippet for producing PPCs and VPCs:

m2s <- m2 %>% child(run_id = "m2s") %>%
  update_parameters(m2) %>%
  convert_to_simulation(subpr = 50) %>%

m2s %>% nm_render("Scripts/basic_vpc.Rmd")
m2s %>% nm_render("Scripts/basic_ppc.Rmd")

Advanced functionality enables groups of runs to be handled with the same concise syntax (no loops):

m1rep <- m1 %>% child(run_id = 1:5) %>% 
  init_theta(init = rnorm(init, mean = init, sd = 0.3)) %>%
  init_omega(init = runif(init, min = init/2, max = init/2)) %>%
  run_in("Models/m1_perturb_inits") %>%

See the website vignette for more examples