rtransparent
is an R package that automatically identifies and extracts indicators of transparency (data availability, code availability, statements of conflicts of interest, statements of funding, statements of protocol registration) from TXT and PMC XML versions of published research articles in biomedicine.
Stylianos (Stelios) Serghiou (sstelios@stanford.edu). This package also utilizes functions from the oddpub
package of Nico Riedel (nico.riedel@bihealth.de).
rtransparent
was validated and subsequently used to extract indicators of transparency from the entire open access literature from PubMed Central. This work is available as a preprint on bioRxiv at: https://www.biorxiv.org/content/10.1101/2020.10.30.361618v1.
First, install the oddpub
package.
devtools::install_github("quest-bih/oddpub")
Then, install the rtransparent
package. Note that you need to use the argument
build_vignettes = T
to render the vignette, otherwise this will not be
accessible.
devtools::install_github("serghiou/rtransparent", build_vignettes = T)
The best way to learn how to use this package is to inspect the vignette. This
can be done by installing the package as explained above and then using the
command vignette("rtransparent")
. This package uses the following naming
convention: functions that work with TXT files extracted from PDF files do not
end in _pmc
. Functions that work with XML files from PubMed Central (PMC) end
in _pmc
. To run all functions on PMC use the rt_all_pmc
function. To get all
meta-data from PMC articles, use the rt_meta_pmc
function.
- Harmonization of code - all code will adopt the naming convention used in
rt_coi_pmc.R
. - Commenting - more comments will be added to explain non-trivial aspects of the code.
- Vignette - a vignette explaining how to use the code is coming.
If you encounter a bug, please file an issue with a minimal reproducible example here and please Label it as a "bug" (option on the right of your window). For help on how to use the package, please file an issue with a clearly actionable question here and label it as "help wanted." For all other questions and discussion, please email the author.