Function to set up a kinetic model with one or more state variables

This function is usually called using a call to mkinsub() for each observed variable, specifying the corresponding submodel as well as outgoing pathways (see examples).

Print mkinmod objects in a way that the user finds his way to get to its components.

mkinmod(
  ...,
  use_of_ff = "max",
  name = NULL,
  speclist = NULL,
  quiet = FALSE,
  verbose = FALSE,
  dll_dir = NULL,
  unload = FALSE,
  overwrite = FALSE
)

# S3 method for mkinmod
print(x, ...)

mkinsub(submodel, to = NULL, sink = TRUE, full_name = NA)

Arguments

...: For each observed variable, a list as obtained by mkinsub() has to be specified as an argument (see examples). Currently, single first order kinetics "SFO", indeterminate order rate equation kinetics "IORE", or single first order with reversible binding "SFORB" are implemented for all variables, while "FOMC", "DFOP", "HS" and "logistic" can additionally be chosen for the first variable which is assumed to be the source compartment. Additionally, mkinsub() has an argument to, specifying names of variables to which a transfer is to be assumed in the model. If the argument use_of_ff is set to "min" and the model for the compartment is "SFO" or "SFORB", an additional mkinsub() argument can be sink = FALSE, effectively fixing the flux to sink to zero. In print.mkinmod, this argument is currently not used.
use_of_ff: Specification of the use of formation fractions in the model equations and, if applicable, the coefficient matrix. If "max", formation fractions are always used (default). If "min", a minimum use of formation fractions is made, i.e. each first-order pathway to a metabolite has its own rate constant.
name: A name for the model. Should be a valid R object name.
speclist: The specification of the observed variables and their submodel types and pathways can be given as a single list using this argument. Default is NULL.
quiet: Should messages be suppressed?
verbose: If TRUE, passed to inline::cfunction() if applicable to give detailed information about the C function being built.
dll_dir: Directory where an DLL object, if generated internally by inline::cfunction(), should be saved. The DLL will only be stored in a permanent location for use in future sessions, if 'dll_dir' and 'name' are specified. This is helpful if fit objects are cached e.g. by knitr, as the cache remains functional across sessions if the DLL is stored in a user defined location.
unload: If a DLL from the target location in 'dll_dir' is already loaded, should that be unloaded first?
overwrite: If a file exists at the target DLL location in 'dll_dir', should this be overwritten?
x: An mkinmod object.
submodel: Character vector of length one to specify the submodel type. See mkinmod for the list of allowed submodel names.
to: Vector of the names of the state variable to which a transformation shall be included in the model.
sink: Should a pathway to sink be included in the model in addition to the pathways to other state variables?
full_name: An optional name to be used e.g. for plotting fits performed with the model. You can use non-ASCII characters here, but then your R code will not be portable, i.e. may produce unintended plot results on other operating systems or system configurations.

Value

A list of class mkinmod for use with mkinfit(), containing, among others,

diffs: A vector of string representations of differential equations, one for each modelling variable.
map: A list containing named character vectors for each observed variable, specifying the modelling variables by which it is represented.
use_of_ff: The content of use_of_ff is passed on in this list component.
deg_func: If generated, a function containing the solution of the degradation model.
coefmat: The coefficient matrix, if the system of differential equations can be represented by one.
cf: If generated, a compiled function calculating the derivatives as returned by cfunction.

A list for use with mkinmod.

Details

For the definition of model types and their parameters, the equations given in the FOCUS and NAFTA guidance documents are used.

For kinetic models with more than one observed variable, a symbolic solution of the system of differential equations is included in the resulting mkinmod object in some cases, speeding up the solution.

If a C compiler is found by pkgbuild::has_compiler() and there is more than one observed variable in the specification, C code is generated for evaluating the differential equations, compiled using inline::cfunction() and added to the resulting mkinmod object.

Note

The IORE submodel is not well tested for metabolites. When using this model for metabolites, you may want to read the note in the help page to mkinfit.

References

FOCUS (2006) “Guidance Document on Estimating Persistence and Degradation Kinetics from Environmental Fate Studies on Pesticides in EU Registration” Report of the FOCUS Work Group on Degradation Kinetics, EC Document Reference Sanco/10058/2005 version 2.0, 434 pp, http://esdac.jrc.ec.europa.eu/projects/degradation-kinetics

NAFTA Technical Working Group on Pesticides (not dated) Guidance for Evaluating and Calculating Degradation Kinetics in Environmental Media

Author

Johannes Ranke

Examples


# Specify the SFO model (this is not needed any more, as we can now mkinfit("SFO", ...)
SFO <- mkinmod(parent = mkinsub("SFO"))

# One parent compound, one metabolite, both single first order
SFO_SFO <- mkinmod(
  parent = mkinsub("SFO", "m1"),
  m1 = mkinsub("SFO"))
#> Temporary DLL for differentials generated and loaded
print(SFO_SFO)
#> <mkinmod> model generated with
#> Use of formation fractions $use_of_ff: max 
#> Specification $spec:
#> $parent
#> $type: SFO; $to: m1; $sink: TRUE
#> $m1
#> $type: SFO; $sink: TRUE
#> Coefficient matrix $coefmat available
#> Compiled model $cf available
#> Differential equations:
#> d_parent/dt = - k_parent * parent
#> d_m1/dt = + f_parent_to_m1 * k_parent * parent - k_m1 * m1

# \dontrun{
 fit_sfo_sfo <- mkinfit(SFO_SFO, FOCUS_2006_D, quiet = TRUE, solution_type = "deSolve")
#> Warning: Observations with value of zero were removed from the data

 # Now supplying compound names used for plotting, and write to user defined location
 # We need to choose a path outside the session tempdir because this gets removed
 DLL_dir <- "~/.local/share/mkin"
 if (!dir.exists(DLL_dir)) dir.create(DLL_dir)
 SFO_SFO.2 <- mkinmod(
   parent = mkinsub("SFO", "m1", full_name = "Test compound"),
   m1 = mkinsub("SFO", full_name = "Metabolite M1"),
   name = "SFO_SFO", dll_dir = DLL_dir, unload = TRUE, overwrite = TRUE)
#> Temporary DLL for differentials generated and loaded
#> Copied DLL from /tmp/Rtmpo7fLwX/file774c3752c3c97.so to /home/jranke/.local/share/mkin/SFO_SFO.so
# Now we can save the model and restore it in a new session
saveRDS(SFO_SFO.2, file = "~/SFO_SFO.rds")
# Terminate the R session here if you would like to check, and then do
library(mkin)
SFO_SFO.3 <- readRDS("~/SFO_SFO.rds")
fit_sfo_sfo <- mkinfit(SFO_SFO.3, FOCUS_2006_D, quiet = TRUE, solution_type = "deSolve")
#> Warning: Observations with value of zero were removed from the data

# Show details of creating the C function
SFO_SFO <- mkinmod(
  parent = mkinsub("SFO", "m1"),
  m1 = mkinsub("SFO"), verbose = TRUE)
#> Program source:
#>   1: #include <R.h>
#>   2: 
#>   3: 
#>   4: static double parms [3];
#>   5: #define k_parent parms[0]
#>   6: #define f_parent_to_m1 parms[1]
#>   7: #define k_m1 parms[2]
#>   8: 
#>   9: void initpar(void (* odeparms)(int *, double *)) {
#>  10:     int N = 3;
#>  11:     odeparms(&N, parms);
#>  12: }
#>  13: 
#>  14: 
#>  15: void diffs ( int * n, double * t, double * y, double * f, double * rpar, int * ipar ) {
#>  16: 
#>  17: f[0] = - k_parent * y[0];
#>  18: f[1] = + f_parent_to_m1 * k_parent * y[0] - k_m1 * y[1];
#>  19: }
#> Temporary DLL for differentials generated and loaded

# The symbolic solution which is available in this case is not
# made for human reading but for speed of computation
SFO_SFO$deg_func
#> function (observed, odeini, odeparms) 
#> {
#>     predicted <- numeric(0)
#>     with(as.list(odeparms), {
#>         t <- observed[observed$name == "parent", "time"]
#>         predicted <<- c(predicted, SFO.solution(t, odeini["parent"], 
#>             k_parent))
#>         t <- observed[observed$name == "m1", "time"]
#>         predicted <<- c(predicted, (((k_m1 - k_parent) * odeini["m1"] - 
#>             f_parent_to_m1 * k_parent * odeini["parent"]) * exp(-k_m1 * 
#>             t) + f_parent_to_m1 * k_parent * odeini["parent"] * 
#>             exp(-k_parent * t))/(k_m1 - k_parent))
#>     })
#>     return(predicted)
#> }
#> <environment: 0x55556118b9d0>

# If we have several parallel metabolites
# (compare tests/testthat/test_synthetic_data_for_UBA_2014.R)
m_synth_DFOP_par <- mkinmod(
 parent = mkinsub("DFOP", c("M1", "M2")),
 M1 = mkinsub("SFO"),
 M2 = mkinsub("SFO"),
 quiet = TRUE)

fit_DFOP_par_c <- mkinfit(m_synth_DFOP_par,
  synthetic_data_for_UBA_2014[[12]]$data,
  quiet = TRUE)
# }