mighty.metadata mighty.metadata website

Overview

Clinical data scientists programming CDISC Analysis Data Model (ADaM) datasets need specifications that are both human-readable and machine-readable. These specifications typically live in Excel spreadsheets that are hard to version-control, difficult to validate, and cannot be consumed by automation tooling. mighty.metadata replaces that workflow with YAML files.

The package loads YAML specifications, validates them against a JSON schema, and provides a pipe-friendly R interface for inspecting and modifying column, parameter, and row definitions. It works at two levels: a single ADaM domain (mighty_domain()) or an entire study (mighty_study()).

mighty.metadata is part of the mighty framework for automated ADaM programming. It is built on S7 and S7schema for robust validation and modern OOP.

Installation

# Install from CRAN:
install.packages("mighty.metadata")

# Install the development version from GitHub:
# pak::pak("NovoNordisk-OpenSource/mighty.metadata")
install.packages("mighty.metadata", repos = c('https://novonordisk-opensource.r-universe.dev', 'https://cloud.r-project.org'))

Usage

Load and inspect a single domain

mighty_domain() loads a YAML metadata file for a single ADaM dataset and validates it against the schema on load.

library(mighty.metadata)

advs <- mighty_domain(
  file = system.file("examples", "advs.yml", package = "mighty.metadata")
)

advs
#> <mighty.metadata::mighty_domain>
#> ADVS: Vital Signs Analysis Dataset
#> Class: BASIC DATA STRUCTURE
#> Keys: USUBJID, PARAMCD, and AVISITN

Use list_columns(), list_parameters(), and list_rows() to see what the specification contains.

list_columns(advs)
#>  [1] "STUDYID"  "USUBJID"  "SAFFL"    "TRTP"     "VISITNUM" "AVISITN" 
#>  [7] "AVISIT"   "PARAMCD"  "PARAM"    "AVAL"     "AVALC"

list_parameters(advs)
#> [1] "BMI"    "BMIGRP"

list_rows(advs)
#> [1] "BASELINE"

Modify a domain

Each dimension (columns, parameters, rows) has a consistent set of verbs: list_*, add_*, remove_*, update_*, select_*, and move_*. These return a modified mighty_domain object and work naturally with the pipe.

# Add a new column
advs |>
  add_column(id = "TRTA", label = "Actual Treatment", .pos = 3) |>
  list_columns()
#>  [1] "STUDYID"  "USUBJID"  "TRTA"     "SAFFL"    "TRTP"     "VISITNUM"
#>  [7] "AVISITN"  "AVISIT"   "PARAMCD"  "PARAM"    "AVAL"     "AVALC"

# Remove columns
advs |>
  remove_columns(c("VISITNUM", "AVALC")) |>
  list_columns()
#> [1] "STUDYID" "USUBJID" "SAFFL"   "TRTP"    "AVISITN" "AVISIT"  "PARAMCD"
#> [8] "PARAM"   "AVAL"

Work with a full study

mighty_study() loads every YAML file from a directory. Each file becomes a named mighty_domain. The optional _study.yml provides study-level properties as a study_config and _mighty.yml provides mighty framework configuration as a mighty_config.

study <- mighty_study(
  path = system.file("examples", package = "mighty.metadata")
)

study
#> <mighty.metadata::mighty_study/list/S7_object>
#> @ mighty: <mighty.metadata::mighty_config>
#> @ study: <mighty.metadata::study_config>
#> $ ADAE: <mighty.metadata::mighty_domain>
#> $ ADSL: <mighty.metadata::mighty_domain>
#> $ ADVS: <mighty.metadata::mighty_domain>

names(study)
#> [1] "ADAE" "ADSL" "ADVS"

study@study
#> <mighty.metadata::study_config>
#> Study ID: example_study
#> Fields: `study_id`
study@mighty
#> <mighty.metadata::mighty_config>
#> External data: 3 sources (`DM`, `VS`, and `AE`)
#> Repos: 2 (`NovoNordisk-OpenSource/mighty.standards/components@main` and `.`)

Useful links