Reproducible Clinical Trial Data and Safety Monitoring Board (DSMB) Reporting with R and tangram

This project is a best practices template for constructing a complex reporting task, in this case a DSMB report. The data pipeline is carefully constructed to be restartable from any point, and it creates a clear separation of concerns. Spreading the data pipeline code throughout a report is bad practice and generally results in copy and paste errors and tables that don't agree. By utilizing this data pipeline strategy and making a small investment in structuring your project, one can save a lot of frustration and effort in the bigger scope of the project.

• data/YYYYMMDD-HHMM contains all data from a given date (raw and transformed).
• pipeline contains data pipeline processing code. This pipeline is restartable from any script in the chain. It continues execution till the end of the chain by design.

• dsmb-open-example.Rmd is the example Rmarkdown for an open report.
• dsmb-closed-example.Rmd is the example Rmarkdown for a closed report.
• dsmb-report.template is the main LaTeX shell for the report.
• pipeline/01-extract.R Extract raw data script.
• pipeline/02-save-raw.R Save the raw data asis for reproducibility.
• pipeline/03-load-raw.R Load the raw data into memory.
• pipeline/04-patch.R Apply any necessary raw data patches.
• pipeline/05-transform.R Transform the raw data into something easily used in reporting.
• pipeline/06-save.R Save the transformed data in a useful format.
• pipeline/07-load.R Load the transformed data into memory. This is the only script required to compile a report if the pipeline is complete.
• pipeline/current-timestamp.txt The timestamp of which directory to use in the data directory for working with data. This is overwritten by 02-save-raw.R when a new extraction is done.