Auditing Studies Across Several Servers

In dsROCrate: 'DataSHIELD' RO-Crate Governance Functions

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  out.width = "100%"
)

SERVER <- "https://opal-demo.obiba.org"

# The examples need OBiBa's public Opal demo server. If it cannot be reached
# (e.g., when building offline), the code is shown but not run.
demo_available <- local({
  old <- options(timeout = 5)
  on.exit(options(old))
  tryCatch(
    {
      con <- url(SERVER, open = "rb")
      close(con)
      TRUE
    },
    error = function(e) FALSE,
    warning = function(w) FALSE
  )
})
knitr::opts_chunk$set(eval = demo_available)

A federated study usually spans several DataSHIELD servers (sites). Both audit() and report() accept a named list of connections, one per site, and return one audit per element plus a combined report.

If you have not used audit() and report() before, start with vignette("getting-started", package = "dsROCrate").

library(dsROCrate)

1. Connect to each site

In a real study, each connection points to a different server. To keep this example self-contained, we open two connections to the same demo server and name them as if they were two sites. Use an administrator or Audit system account at every site.

USERNAME <- "administrator"
USERPASS <- "password"

cons <- list(
  site_a = opalr::opal.login(USERNAME, USERPASS, url = SERVER),
  site_b = opalr::opal.login(USERNAME, USERPASS, url = SERVER)
)

2. Audit the study

The arguments of audit() are applied to every site, so the project (and, if given, the user and tables) must exist in each of them.

study_audit <- cons |>
  audit(project = "CNSIM", tables = "CNSIM1", user = "dsuser")

# one RO-Crate per site
names(study_audit)

print(study_audit$site_a)

3. Create a study report

As for a single crate, we can first generate only the contents of the report:

report_file <- tempfile(fileext = ".Rmd")

study_parts <- study_audit |>
  report(filepath = report_file, render = FALSE)

# overview diagram
study_parts$overview_diagram

# overview table (Safe People, Safe Project and Safe Data)
knitr::kable(study_parts$overview_data)

and render the full report (HTML) with:

study_audit |>
  report(
    filepath = report_file,
    title = "DataSHIELD study audit",
    render = TRUE,
    overwrite = TRUE
  )

4. Close the connections

invisible(lapply(cons, opalr::opal.logout))

unlink(report_file, force = TRUE)

dsROCrate documentation built on Oct. 16, 2026, 5:09 p.m.