In MikeKSmith/rspeaksnonmem: Read, Modify and run NONMEM from within R

 devtools::load_all(pkg = ".")
file.copy(from = file.path(system.file("exdata", package = "rspeaksnonmem"),
                           "warfarin_conc_pca.csv"),
            to = getwd(), overwrite = T )
file.copy(from = file.path(system.file("exdata", package = "rspeaksnonmem"),
                           "warfarin.ctl"),
            to = getwd(), overwrite = T )
file.rename(from = "warfarin.ctl", "run0.ctl")

In this notebook we will look at running NONMEM and PsN from R using the R package rspeaksnonmem.

Overview

This demo will show the following steps:

• Running NONMEM and PsN
• Integrating results into a Notebook.

Step 1: Running NONMEM

First, let's run a model and knit the results back into this notebook. Note that in the command argument to the function we specify the complete path to the NONMEM executable .bat file. This allows you to be very precise about which version of NONMEM you use. If the clean argument is non-zero it cleans up the working directory deleting all of the temporary NONMEM files, leaving only the output.

  estimate_NM(command = "c:/nm72/run/nmfe72.bat",
              modelFile = "run0.ctl",
              clean = 1)

Step 2: Summarise estimation:

We can now use PsN command sumo to summarise the output. Note that here we are not using the command argument, but instead we specify the installation directory for Perl / PsN and which PsN version we are using. rspeaksnonmem will then create a command based on the directory, sumo and version information.

  sumo_PsN(installPath = "c:/strawberry/perl",
           version = "4.7.0",
           lstFile = "run0.lst")

Step 3: Model Diagnostics

Next, we might want to produce some model diagnostics using the R package xpose4. To do this, we need to create an Xpose database object and then create the diagnostic plots. Xpose needs the run number in order to pick up the $TABLE file outputs from NONMEM.

  runno <- as.numeric(gsub("[a-z]", "", list.files(pattern = "^sdtab")[1]))

  ## ----createXpdb----------------------------------------------------------
  xpdb <- xpose4::xpose.data(runno, quiet = T)
  # save(base.xpdb, file='Xpose database.RData')

DV vs PRED / IPRED

  print(xpose4::dv.vs.pred.ipred(xpdb))

PRED vs IDV

  print(xpose4::pred.vs.idv(xpdb))

PRED vs IDV

  print(xpose4::ipred.vs.idv(xpdb))

WRES vs IDV

  print(xpose4::wres.vs.idv(xpdb))

WRES vs PRED

  print(xpose4::wres.vs.pred(xpdb))

EBEs histogram

  print(xpose4::ranpar.hist(xpdb))

Individual Plots

  print(xpose4::ind.plots(xpdb, layout = c(4, 4)))  

Step 4: VPC diagnostisc

We may also want to investigate simulation-based diagnostics using the PsN routine VPC. In the call to VPC_PsN we have specified the required arguments for VPC (samples, seed) but you can also pass additional arguments to the VPC routine by passing a name-value list to the argument psnOpts. rspeaksnonmem will check these arguments against the options for VPC and let you know if any are mis-specified. Here the argument min_points_per_bin is not a valid option for VPC. (The correct option is min_points_**in**_bin.)

VPC_PsN( modelFile='run0.ctl', 
         installPath = "c:/strawberry/perl",
         version = "4.7.0",
         working.dir='Run1_VPC', 
         samples=100, seed='54321',
         psnOpts=list(lst='run0.lst', 
                      auto_bin = '5',
                      min_points_per_bin = '2'))

and then use the Xpose function xpose_vpc to plot the result:

vpcdir <- "Run1_VPC"

vpctab <- file.path(vpcdir,"vpctab")
vpcinfo <- file.path(vpcdir,"vpc_results.csv")

xpose4::xpose.VPC(vpc.info=vpcinfo, vpctab=vpctab)

MikeKSmith/rspeaksnonmem documentation built on March 12, 2023, 3:25 p.m.