tcp-lab/r4tcpl

A collection of utility R functions used in a variety of our projects.

r4tcpl

A collection of utility functions used in our R projects

Packages are the fundamental units of reproducible R code. They include reusable R functions, the documentation that describes how to use them, and sample data. You can find a thorough discussion of this topic in: R packages (2e) by Hadley Wickham and Jenny Bryan, available online at: https://r-pkgs.org/

Building dependencies

This is what you need to build the package, not the packages that r4tcpl will need to run properly. Those are listed further down (see Package dependencies section).

1. If you are working on Windows, install the RTools version appropriate for the R release you are using.

2. Make sure devtools, roxygen2, testthat, and knitr packages are already installed, otherwise install them running: install.packages(c("roxygen2", "testthat", "knitr", "devtools")).

Basic workflow to add new functions and data

1. Write/copy your function(s) into a .R file within .../r4tcpl/R/.

2. Remember to include the @export tag in the roxygen comments of all the functions you want to make visible when r4tcpl package will be loaded. When building the package, the corresponding export() statement will be automatically generated inside NAMESPACE file.

3. It’s important to note that library() should NEVER be used inside a package. Use instead the :: notation to explicitly refer specific namespaces.

4. For each dependency (i.e., wherever you used the syntax <package_name>::<object>) remember to add package_name to Imports field within the DESCRIPTION file by typing usethis::use_package("package_name", min_version=TRUE) from the project directory .../r4tcpl/ (set it as the working directory in RStudio).

5. Unlike User Libraries, System Libraries don't need to be declared as dependencies (since they are expected to be always installed by default in every R distribution). However, whenever some function from a System Library other than The R Base Package (base) is used it is recommended to import it or the entire library in the package namespace by adding the roxygen tags @importFrom or @import, respectively. When building the package, the corresponding import() statement will be automatically generated inside NAMESPACE file.

6. Add example data sets to the package running usethis::use_data(<my_pkg_data>) from the project directory. This command will save the data contained in the R variable <my_pkg_data> to the folder .../r4tcpl/data/ as a binary .rda representation (storing one R object in each .rda file).

7. Remember to provide a complete documentation for both functions (by roxygen2 heading) and data (by editing the file .../r4tcpl/R/data.R), otherwise you will get a warning when building the package.

8. From the project directory, run devtools::document() to convert roxygen comments into proper R documentation and (re)generate NAMESPACE (based on @export tags). Then you can use pkgload::dev_help('<function_name>') to have a quick preview of the <function_name>.Rd file.

9. Possibly update tests in .../r4tcpl/tests/testthat/test-data_exploration.R.

10. Possibly update the package version in the DESCRIPTION file

11. From the project directory, run devtools::check() to build the package (remember that all @examples will be used as tests in this phase; to prevent some of them from being run put them inside a \dontrun{} section).

12. Use devtools::load_all() to load the package from the local directory and possibly test the new features.

13. When you are happy, you can git add, commit, and push.

14. Regardless of your wd, use devtools::install_github("TCP-Lab/r4tcpl") to install (or update) the package from GitHub, and load it as usual through library(r4tcpl).

Package dependencies

The following packages need to be preinstalled before running devtools::install_github("TCP-Lab/r4tcpl").

• GEOquery
• ggplot2
• svDialogs
• tools
• dplyr
• stringr
• AnnotationDbi
• VennDiagram
• futile.logger
• org.Hs.eg.db
• mclust

How to rename the package

1. Update the package name field in the DESCRIPTION file.
2. Update all package names in the README file.
3. Update the .RProj name.
4. Update the .Rbuildignore file.
5. Within the folders .../r4tcpl/R/ and .../r4tcpl/tests/ look for mentions of the old package name within all .R files and update them.
6. git add, commit, and push these changes to the old repository.
7. Delete the local repository folder.
8. Go to GitHub and use the repository Settings to rename it.
9. git clone locally the renamed repository.
10. From RStudio, run devtools::document() to update documentation (i.e., Rd files).
11. Run devtools::check() to build the renamed package.
12. If everything went fine, git add, commit, and push some changes to the new repository as a final test.
13. Finally, remove your old package from R by remove.packages("oldName")...
14. ...and install the new one devtools::install_github("TCP-Lab/newName")

r4tcpl's Issues

In `lms()`, `cols` and `rows` options are not effective for list elements

lms() arguments cols and rows are ignored when the function is applied recursively to list elements. Just propagate the option values to the recursive call of lms().