NMreadExt: Read information from Nonmem ext files

In NMdata: Preparation, Checking and Post-Processing Data for PK/PD Modeling

Read information from Nonmem ext files

Description

Read information from Nonmem ext files

Usage

NMreadExt(
  file,
  return,
  as.fun,
  modelname,
  col.model,
  auto.ext,
  tableno = "max",
  file.ext,
  slow
)

Arguments

file	Path to the ext file
return	The .ext file contains both final parameter estimates and iterations of the estimates. If return="pars" (default) the final estimates are returned in addition to what other parameter-level information is found, like FIX, sd etc. as columns. If return="iterations", the iterations are returned (including objective function value). If return="obj" objective function value at final estimate is returned. If return="all", all er returned, though in separate data.frames compiled in a list.
as.fun	The default is to return data as a data.frame. Pass a function in as.fun to convert to something else. If 'data.table's are wanted, use 'as.fun="data.table"'. The default can be configured using NMdataConf().
modelname	See '?NMscanData'
col.model	See '?NMscanData'
auto.ext	If 'TRUE' (default) the extension will automatically be modified using NMdataConf()$file.ext. This means 'file' can be the path to an input or output control stream, and NMreadExt will still read the '.ext' file.
tableno	In case the ext file contains multiple tables, this argument controls which one to choose. The options are "max" (default) Pick the table with the highest table number. This typically means the results from the last '$ESTIMATION' step are used. "min" Pick results from the first table available. "all" Keep all results. The tables can be distinguished by the tableno column. an integer greater than 0, in which case the table with this table number will be picked.
file.ext	Deprecated. Please use file instead.
slow	Use a slow but more robust method to read tables? If missing or 'NULL', the fast method will be tried first, and if any issues are seen, the method will switch to 'slow=TRUE'. If 'FALSE',it will also switch in case of issues, but a warning is issued. In other words, it should be safe to not use this argument.

Details

The parameter table returned if return="pars" or return="all" will contain columns based on the Nonmem 7.5 manual. It defines codes for different parameter-level values. They are:

-1e+09: se -1000000002: eigCor -1000000003: cond -1000000004: stdDevCor -1000000005: seStdDevCor -1000000006: FIX -1000000007: termStat -1000000008: partLik

The parameter name is in the parameter column. The "parameter type", like "THETA", "OMEGA", "SIGMA" are available in the par.type column. Counters are available in i and j columns. j will be NA for par.type=="THETA"

The objective function value is included as a parameter.

Notice that in case multiple tables are available in the 'ext' file, the column names are taken from the first table. E.g., in case of SAEM/IMP estimation, the objective function values will be in the SAEMOBJ column, even for the IMP step. This may change in the future.

Value

If return="all", a list with a final parameter table and a table of the iterations. If return="pars", only the parameter table, and if return="iterations" only the iterations table. If you need both, it may be more efficient to only read the file once and use return="all". Often, only one of the two are needed, and it more convenient to just extract one.

NMdata documentation built on Nov. 5, 2025, 5:43 p.m.