Troubleshooting

Start with the exact error, package version and loaded path. Retrying with different scientific settings can change the question without solving the problem.

Repository access is denied

Check Git availability and whether your GitHub account can access IOBR/IOBRcausal. Use your normal Git credential manager. Keep tokens out of commands saved in scripts or reports. A private repository may return 404 to an anonymous client.

R loads an older package

packageVersion("IOBRcausal")
find.package("IOBRcausal")
.libPaths()

Restart R after installing an update. Check child Rscript processes as well as the interactive session when multiple libraries exist.

The terminal cannot find iobrcausal

The package launcher can be called from R without adding it to PATH:

launcher <- system.file("exec", "iobrcausal.R", package = "IOBRcausal")
rscript <- file.path(R.home("bin"),
  if (.Platform$OS.type == "windows") "Rscript.exe" else "Rscript")
system2(rscript, c(shQuote(launcher), "start"))

For Skill installation, also check node --version, npm --version and git --version. The documented standalone command requires Node.js 22.20 or later.

A method was skipped

Read the method-specific reason. Typical causes are an unsupported exposure/outcome/target combination, insufficient support or a missing optional dependency. A skip is different from a fitted null effect or an attempted failure.

No candidates were retained

Check the requested family, missingness, sample/event counts, gene-set coverage, configured gates and full audit tables. A completed workflow with no retained candidates is a valid result. Keep it distinct from a failed or incomplete stage.

Excel or a figure is missing

Inspect the actual artifact list, writer dependency status and validation output. Look for an explicitly reported CSV fallback. Saved-number plots need matching numeric evidence; a survival curve needs a saved curve cache.