As an open source software system, R is continuously evolving. In the clinical trials, reliable and valid software is crucial and hence enforced by regulatory authorities (FDA 2003). In order to establish proof of the package’s validity on our server, we developed a validation suite following the PHUSE R Package Validation Framework (Hughes 2021). We first formulated user requirements. For each requirement, human-readable test cases were written, i.e. small scenarios for checking the package’s functionality. Subsequently, an independent programmer (none of the package developers) implemented the test cases in code, reporting bugs and unclear documentation back to the developers.
The whole validation suite was run on our server, and the validation
report can be found here.
As every system configuration is different, we can not guarantee that
the drugdevelopR
package will work as expected on your
system. The validation suite may serve as a building block in your own
validation framework. In the following, we will explain how to use
it.
The validation report is compiled using our fork of PHUSE’s valtools
package. We recommend using our forke where a few bugs were fixed and
features were added. This changes were not yet merged into to the
package’s main repository. It can be installed by running the following
command:
In the next step, download the source
files of the current drugdevelopR
project. Extract
the folder and navigate to it using RStudio. Install the package from
source using the following command:
Then, open the file ./validation/validation.Rmd
in
RStudio and use the Knit button for compiling the report. This
may take a long time due to the long runtime of the test codes. We used
12 cores of our RStudio server, which took 93.5 hours (i.e. almost four
days) to complete.
If you need a signature page within your report, proceed as follows:
./validation/validation.yml
and replace
our names by the names of the responsible persons at your facility../validation/validation.Rmd
and uncomment
the section ## Certifications {-}
and its subsection
(approx. 20 lines of code).FDA, US Food and Drug Administration (2003): Guidance for Industry Part 11. https://www.fda.gov/media/75414/download. Ellis Hughes and the PHUSE Data Visualisation & Open Source Technology Working Group (2021): White Paper 059. R Package Validation Framework. https://advance.phuse.global/display/WEL/R+Package+Validation+Framework.