TAF Checks

TAF complete analyses are characterized by the following criteria:

1. The main analysis folder must contain a boot folder. For older analyses, this folder may be named bootstrap.

2. The boot folder must contain a DATA.bib file, declaring the initial data used as a starting point in the analysis. The boot folder may also contain a SOFTWARE.bib file.

3. After running the TAF function taf.boot(), data files must be present in the boot/data folder, corresponding to the declarations in the DATA.bib file. Similarly, after running taf.boot(), software files must be present in the boot/software folder if a SOFTWARE.bib file exists.

4. If files exist both in boot/initial/data and the boot/data folder, then the file contents should be identical.

5. Files and folders inside the boot/data folder must be declared in the DATA.bib file. Similarly, files and folders inside the boot/software folder must be declared in the SOFTWARE.bib file.

6. The main analysis folder should contain TAF scripts named data.R, model.R, output.R, and report.R. These scripts may call other R scripts and/or dynamic reports with file extensions such as *.qmd, *.Rmd, or *.rmd. For some analyses, the model script may be named method.R.

7. R code should use relative paths rather than absolute paths.

8. After running the TAF function source.all(), data files should be present in the data folder and output files should be present in the output folder. The model folder may contain intermediate results, and the report folder may contain final formatted results. For some analyses, the model folder may be named method.

9. To make it easy for the scientific community to browse and review the main data and results from the analysis, the file formats CSV (*.csv for tables), PNG (*.png for raster images), and PDF (*.pdf for vector images) should be used. These file formats are easy to open on any computer and are also easy to view on GitHub.

The above criteria can be evaluated by examining files and folders without running or modifying any part of the analysis. In contrast, the following criteria (marked with *) are best evaluated by running the full analysis, which can take a long time to run and may require special software or user authorization, and in some edge cases make irreversible changes to files.

10.* Regardless of which data and result files are stored online on GitHub, it should be possible to clone the analysis to a local computer and perform a full clean before rerunning the analysis successfully. A full clean consists of running the TAF functions clean.boot(force=TRUE) for the boot folder and clean() for the folders produced by the TAF scripts. A successful rerun of the analysis consists of running taf.boot() and source.all() without errors, producing the same or similar results as the original analysis.

11.* The TAF scripts (data.R, model.R, output.R, report.R) should run sequentially, with each script starting by reading files from a previous step and ending by writing out files.

12.* The data.R script should create the data folder and write files into that folder. Likewise, the model.R, output.R, and report.R scripts should create and write into the corresponding folders.