Working with SO files in R with libsoc

In libsoc: Read, Create and Write 'PharmML' Standard Output (so) XML Files

Reading an SO file

It is easy to read a Standard Output file into R.

filename <- system.file("extdata", "pheno.SO.xml",  package="libsoc")
so <- libsoc::so_SO_read(filename)

Navigating the SO structure

The SO R object closely follows the tree structure of the SO XML. We can extract the population parameter estimates like this:

param <- so$SOBlock[[1]]$Estimation$PopulationEstimates$MLE
param

The result is a data.frame with the population estimates and the names of the parameters as column names. Note the way we go deeper in the SO tree with successive use of the $-opterator. As an SO can contain more than one SOBlock the SOBlock-structure is a list. In the example we simply select the first (and only) SOBlock.

libsoc also has a special function for extracting the parameter estimates across all SOBlocks

so$all_population_estimates()

Getting information on parameters

If the SO is linked to a PharmML (for example if it was generated with "nmoutput2so -generate_pharmml") parameter information can be retrieved from the ParameterModel.

cols <- colnames(param)
so$variability_type(cols)

It is also possible to get information on which parameters represent correlations.

so$correlation_parameters(cols)

This example has no correlation parameters.

The names of the random variables corresponding to the variability parameters can be retrieved:

so$random_variable_from_variability_parameter(cols)

Obtaining some parameter uncertainty measures

The covariance matrix is of standard R matrix type:

so$SOBlock[[1]]$Estimation$PrecisionPopulationEstimates$MLE$CovarianceMatrix

The standard error for the parameters can be extracted for each SOBlock separately:

so$SOBlock[[1]]$Estimation$PrecisionPopulationEstimates$MLE$StandardError

Or more conveniently for all SOBlocks together:

so$all_standard_errors()

Relative standard errors can currently only be retrieved for each SOBlock separately:

so$SOBlock[[1]]$Estimation$PrecisionPopulationEstimates$MLE$RelativeStandardError

Getting the OFV

so$SOBlock[[1]]$Estimation$OFMeasures$Deviance

Getting the Predictions

pred <- so$SOBlock[[1]]$Estimation$Predictions
head(pred, 20)

Getting information on table columns

Each table in the SO structure has a columnType attribute that gives an array of columnTypes taken directly from the xml.

attributes(pred)$columnType

The portable way of finding the column name/number for id, idv and dv columns is to use the provided functions.

libsoc::id_column(pred)

libsoc::id_column_name(pred)

libsoc::idv_column(pred)

libsoc::idv_column_name(pred)

Getting the Residuals

res <- so$SOBlock[[1]]$Estimation$Residuals$ResidualTable
head(res, 20)

Reading messages

Messages from a run is stored in the TaskInformation structure. Each message contains a Name a Severity from 1 to 10, the message content and the name of the tool that emitted the message.

messages <- so$SOBlock[[1]]$TaskInformation$Message
messages[[1]]$Name
messages[[1]]$Content
messages[[1]]$Severity
messages[[1]]$Toolname
messages[[13]]$Content

iOFV

OFV values for each individual separately is in the OFMeasures section

iofv <- so$SOBlock[[1]]$Estimation$OFMeasures$IndividualContribToLL
head(iofv)

libsoc documentation built on Feb. 3, 2022, 5:07 p.m.