Getting Started with mrgda

Source: vignettes/getting-started.Rmd

Introduction

This vignette takes the user through some basic scenarios for exploring, assembling, and validating data sets using mrgda, introducing you to its standard workflow and functionality. For more information how to effectively integrate mrgda into your workflow, visit the MeRGE Expo.

Source data

We begin by pointing to a source directory containing the source data (typically STDM or ADaM datasets). read_src_dir will read in every data file and return a named list of the data objects, as well as some additional metadata.

src_path <- system.file("example-sdtm", package = "mrgda")

# Read in all XPT files in source directory
src_list <- mrgda::read_src_dir(src_path, .file_types = "xpt")

Exploring the source data

While it is common practice to use View() for inspecting dataframes or matrix-like R objects, mrgda provides a convenient tool for exploring your entire source directory.

mrgda::v(src_list)

v() facilitates manual inspections, such as confirming data integrity for a specific subject or over specified date range. Note that v() runs a shiny app in the Rstudio viewer pane as background process, allowing you to retain control over your R console:

Some features include:

  • Allows you to seamlessly transition between any of the datasets in your list (you can also pass in a single dataframe)
  • Apply global subject filters, filtering each of the datasets in src_list to only contain the specified subject ID
  • Apply column filters per dataset
  • Groups the data by the specified or detected .subject_col, helping to differentiate between subject records.
  • Column Organization
    • “Freeze” or lock columns in place while scrolling horizontally
    • Relocate columns to another position
  • Other formatting options related to readability and presentation of column label attributes

Query source list

In addition to the convenient viewing and filtering methods provided by v(), you can also query the entire source directory for a specific character string via query_src_list():

# Summarize source directory
query_src_list(src_list, .string = "RACE")
#> # A tibble: 2 × 3
#>   DOMAIN       COLUMNS                  MATCHING
#>   <chr>        <chr>                    <chr>   
#> 1 lb           LBORRES,LBSTRESC,LBSTNRC RACE    
#> 2 mrgda_labels COLUMN_NAME,COLUMN_LABEL RACE

Save assembled data

After you finish assembling your dataset, you can export the compiled data using write_derived, ensuring both convenience and reproducibility. This function produces a CSV, an xpt file, and a dedicated folder housing valuable metadata. This dual export capability enhances data portability, accommodating various use cases. The specification file, nm_spec, is meant to contain information relating to units, abbreviations, decoding options, and more (see the MeRGE Expo for more information on this).

nm_spec <- yspec::ys_load(system.file("derived", "pk.yml", package = "mrgda"))
mrgda::write_derived(
  .data = derived$nm,
  .spec = nm_spec,
  .file = rprojroot::find_rstudio_root_file("data", "derived", "pk.csv")
)

Other functions

mrgda exports other helpful functions for data creation, management, and validation. See the Reference Log for more details.