Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/namakala/inappropriate-acid-suppressor-agent-use/llms.txt

Use this file to discover all available pages before exploring further.

Once your environment is set up and data/raw/data.csv is in place, you interact with the pipeline entirely through the targets package. targets tracks every input, function, and output as a node in a dependency graph, so you can run the full analysis, check what is out of date, load individual results, and debug failures — all without manually tracking which scripts to re-run.

​
Running the pipeline

Call tar_make() from your R console (or in an RStudio terminal) to execute the full pipeline: 
targets::tar_make()
targets reads _targets.R, resolves the dependency graph, and dispatches targets to up to four parallel workers via crew. Targets whose inputs have not changed since the last run are skipped automatically. A green checkmark next to a target name means it was skipped as up to date; an orange label means it was dispatched and executed.
Run targets::tar_make() from the project root directory, where _targets.R is located. If you are in a subdirectory, targets will not find the pipeline definition.

​
Checking pipeline status

Use these commands to inspect the state of the pipeline before or after a run.

​
Visual dependency graph

targets::tar_visnetwork()
Opens an interactive network diagram in your browser showing every target, its status (up to date, outdated, errored), and the edges between them. Use this to understand what will run before committing to a full tar_make().

​
List all targets

targets::tar_manifest()
Returns a data frame listing every target name, the command used to build it, and its configured options. Useful for auditing the pipeline definition programmatically.

​
See which targets need to run

targets::tar_outdated()
Prints the names of all targets that are outdated relative to their current dependencies. A target is outdated if its command, any of its upstream targets, or any of its tracked files have changed since it last ran.

​
Check run progress

targets::tar_progress()
Returns a data frame with the status of every target from the most recent tar_make() call. Status values include "built", "skipped", "errored", "cancelled", and "dispatched".

​
Accessing results

After a successful run, results are stored in _targets/objects/. You can retrieve them without re-running any targets.

​
Load a target into your environment

targets::tar_load(mod_prop)
Loads the named target directly into your global environment. You can then inspect mod_prop like any other R object.

​
Read a target without loading

targets::tar_read(mod_prop)
Returns the target value without assigning it to the global environment. Useful in scripts or R Markdown/Quarto documents where you want to avoid side effects.

​
List all computed targets

targets::tar_objects()
Returns a character vector of all target names that have been built and stored in _targets/objects/. Combine with tar_load_everything() to load all results at once.

​
Key pipeline targets and what they produce

TargetWhat it produces
tblRaw data frame read from data/raw/data.csv
tbl_cleanCleaned and standardized data frame, ready for modeling
mod_propOverall random-effects meta-analysis model (proportion data)
mod_copas_propCopas selection model fit to mod_prop, adjusting for publication bias
plt_forest_propForest plot PDF saved to docs/figures/meta-analysis-prevalence.pdf
mod_subgroup_AgeSubgroup meta-analysis stratified by patient age
mod_subgroup_ContinentSubgroup meta-analysis stratified by continent
mod_subgroup_SettingSubgroup meta-analysis stratified by clinical setting
mod_subgroup_JBI_ClassificationSubgroup meta-analysis stratified by JBI quality score
mod_subgroup_use_guidelineSubgroup meta-analysis stratified by guideline adherence
mod_metareg_AgeUnivariable meta-regression with age as predictor
mod_metareg_ContinentUnivariable meta-regression with continent as predictor
mod_metareg_SettingUnivariable meta-regression with setting as predictor
mod_metareg_JBI_ClassificationUnivariable meta-regression with JBI score as predictor
mod_metareg_use_guidelineUnivariable meta-regression with guideline use as predictor
mod_metareg_mvMultivariable meta-regression with year, JBI score, guideline use, setting, continent, and sample size
reportCompiled Quarto PDF report from docs/report.qmd
readmeRendered README.qmd, built last (priority = 0)

​
Troubleshooting

​
A target fails but the pipeline keeps running

The pipeline is configured with error = "continue": 
tar_option_set(
  error = "continue",
  ...
)
This means that if one target errors, targets marks it as failed and continues building any downstream targets that do not depend on it. After the run, call tar_progress() to identify which targets errored, then inspect those targets individually.

​
Interactive debugging

By default, tar_make() runs each target in a separate callr subprocess, which makes debugging difficult because you cannot drop into a browser or inspect objects interactively. To disable subprocess isolation: 
targets::tar_make(callr_function = NULL)
With callr_function = NULL, targets run in your current R session. You can set breakpoints, use browser(), or inspect intermediate values directly.

​
Packages seem out of sync

The project uses renv to lock package versions. If you install or update packages and targets behave unexpectedly, check whether your library matches the lockfile: 
renv::status()
If renv::status() reports discrepancies, run renv::restore() to reinstall the locked versions, then restart your R session before calling tar_make() again.
For distributed computing across multiple machines or a high-performance cluster, replace crew_controller_local(worker = 4) in _targets.R with an appropriate crew controller (for example, crew.cluster for SLURM). The same targets::tar_make() call is used regardless of which controller is configured.

Build docs developers (and LLMs) love

Get started for freeTalk to us