Most information on computer ressources at INRAE/MIAT and on usage of linux and software for linux is available in Livret Informatique MIAT. Feel free to report any missing or incorrect information. In addition to this tutorial, the current page gathers basic recommandations on practices, links and tools that I recommend to use for your work with me. They are all written for Linux based (mostly Ubuntu) OS. Again, do not hesitate to report mistakes or missing information.
Between meetings, if you have a question or a problem, I prefer to be contacted my mattermost if these are short questions (in a canal shared with the other supervisors if any) or by emails (also put in CC the other supervisors if any): I will usually find a way to meet you soon to help you solve it (if I can not solve it by electronic means). Be as precise as possible explaining your problem (for computing issues, for instance, copy/paste your command, the error messages or join screenshots).
On Ubuntu, RStudio is installed by downloading the Debian package at this link and by running the following command lines (from the download directory):
sudo dpkg -i ... sudo apt install -f
... is the name of the downloaded file. The first command line usually returns an error that is corrected by the second one. R can be installed in two different ways: from the Ubuntu repository or a direct, fixed version, can be installed from RStudio cloud, which is better suited for reproducible analyses.
What I strongly recommend for every projet is to:
have a fixed version of R installed on your computer (and install the most recent version each time you create a new project) as described on this page;
have an associated R project created with RStudio,
myproject.Rproj (for instance);
profile4R.shreferencing the proper R version with:
in it (adapted to your R version)
source profile4R.sh rstudio myproject.Rproj &
To ensure the maximum reproducibility (and efficiency) of your analyses:
. ├─ communications │ ├── 2018-01-12_workshopXXX │ └── 2018-03-15_SeminaireYYY ├── CR ├── data ├── myproject.Rproj ├── results ├── renv.lock ├── renv └── RLib
raw data files must never be manually edited. Unless said differently, a file is modified by creating a script that documents and performs the data edition and that exports a new dataset;
analyses are all performed in scripts or (better) RMarkdown or quarto files fully commented in English (origin of the data, purpose of the script, interpretation of results), including the date of last modification and a
scripts are run from the
RLib repository and must contain only relative paths to data. Never use
read.table("/home/myname/someproject/data/mydata.csv") but use
read.table("../data/mydata.csv") instead. Another (probably even more elegant) solution is to use the
here package and to write your path from the project root systematically;
Finally, for quick&dirty analyses or package development (or every time you don’t need to control the reproducibility of your analyses), I strongly advise using the official ubuntu release of R along with the package repository provided by r2u:
sudo su wget -q -O- https://cloud.r-project.org/bin/linux/ubuntu/marutter_pubkey.asc \ | tee -a /etc/apt/trusted.gpg.d/cran_ubuntu_key.asc echo "deb [arch=amd64] https://cloud.r-project.org/bin/linux/ubuntu jammy-cran40/" \ > /etc/apt/sources.list.d/cran_r.list apt-key adv --keyserver keyserver.ubuntu.com --recv-keys \ 67C2D66C4B1D4339 51716619E084DAB9 apt update apt install r-base-core
sudo su apt update -qq && apt install --yes --no-install-recommends wget \ ca-certificates gnupg wget -q -O- https://eddelbuettel.github.io/r2u/assets/dirk_eddelbuettel_key.asc \ | tee -a /etc/apt/trusted.gpg.d/cranapt_key.asc echo "deb [arch=amd64] https://r2u.stat.illinois.edu/ubuntu jammy main" \ > /etc/apt/sources.list.d/cranapt.list
CRAN and Bioconductor packages are then simply installed with:
sudo apt install r-cran-mixkernel sudo apt install r-bioc-asics
David’s repository provides an (unofficial) INRAE template for RMarkdown and Quarto files, as well as an INRAE
Elise’s shiny apps (based on
cols4all) is useful to choose your color palette when doing plots.
Ghislain’s presentation gives information about useful ressources for package development.
First download of a project is performed using
git clone ... with
... being the repository URL. Once your repository cloned, you can create a RStudio project (as explained in the previous section) associated to it, which should give you access to a Git menu, with easy-to-use buttons.
Good practices on git include:
not versioning heavy data files (that are not supposed to change; these files can be sent to collaborators using FileSender;
not versioning result files (that can be produced from scripts present in the directory or compiles from a
not versioning binary files (or only the ones that are strictly necessary and are not heavy).
To run your scripts, organize your remote directories as your local directories (sending files with the command line
scp or using the linux tool
gftp that can be installed with
sudo apt install gftp or using git). Do not forget to send your
renv.lock file as well and to use the same R version (all R versions are not necessary available on Genotoul). You will have first to reinstall your
renv environment using the
For writing articles or sometimes for creating slides, posters, …, I am usually using $\LaTeX$ that handles perfectly mathematical formula. I strongly recommend that you maintain a unique global
bib file for your bibliography somewhere. Be careful that BibTeX entries imported from website or with zotero usually need some cleans-up (in particular, BibTeX file are recommended to be written in pure ASCII, while automatic importation are usually encoded in UTF-8, e.g., including accents or some special characters; automatic importations also frequently include non-official fields).