For the ordinary installation, start with Install lisaR. This page keeps the optional dependencies and installation diagnostics. You need R 4.1 or later. RStudio is optional, but the steps below work in its Console.
If you only need to read a portable HTML report, you do not need R. Keep the report folder and its relative files together.
lisaR is installed from its GitHub repository. The installation includes the LISA dictionaries and the synthetic sample project used in Quick Start.
Run this once in R with internet access. BiocManager
installs fgsea from a Bioconductor release that is
compatible with your R version, together with the required CRAN
packages.
install.packages(c("BiocManager", "remotes"))
BiocManager::install(
c("digest", "fgsea", "fs", "ggplot2", "hommel", "jsonlite", "openxlsx", "readr", "yaml"),
update = FALSE,
ask = FALSE
)The required hommel version is at least 1.8; the
installer resolves this from DESCRIPTION. Base and
recommended Imports come with R. Install extra packages only for the
route you intend to use:
| Route | Additional packages |
|---|---|
| Shiny application with background figure jobs | shiny, callr, ps |
| Prepared Riaz project | data.table |
| Gene-level figures and FULL cards | ggrepel, gridExtra,
patchwork |
| Native KEGG maps | AnnotationDbi, KEGGREST, png,
plus org.Hs.eg.db for human or org.Mm.eg.db
for mouse |
| Reconstructing Riaz upstream inputs | data.table, readxl, DESeq2,
AnnotationDbi, org.Hs.eg.db,
SummarizedExperiment |
| Building these HTML vignettes from source | knitr, rmarkdown, plus Pandoc outside
R |
Use the same BiocManager::install() form for these
packages. For example, install data.table before using the
prepared Riaz project. Later, validate_lisa_config() will
report packages needed by the analysis and report products you
selected.
In R or the RStudio Console, run:
This installs the package and its included resources in one step. You do not need to locate a separate source archive or install lisaR again to load its dictionaries. Ordinary installation does not rebuild the vignettes.
The source package includes the LISA dictionaries, category hierarchy and synthetic sample project. It does not include the external MSigDB TERM2GENE membership needed for a real study. Preparing that membership is a separate step in Resources and provenance.
Source and release status are listed at DBM-OlmedaLab/lisaR. The repository is public; no GitHub account or access token is required to view or download it.
Restart R if another lisaR version is already loaded, then run:
library(lisaR)
packageVersion("lisaR")
find.package("lisaR")
system.file("extdata", "dictionaries", package = "lisaR")
system.file("examples", "quick-start", "study.yml", package = "lisaR")packageVersion() identifies the installed package.
find.package() shows the R library from which it was
loaded. If these checks complete, continue with Quick Start. That synthetic project contains
every input needed for a first report and does not require MSigDB
access.
For a real study, continue to Prepare differential-expression inputs, then run lisaR with your prepared files.
The R/RStudio route above is complete. A coding assistant is optional and does not remove the need for a suitable local R installation or the required R packages.
Codex CLI can read local files and run tools installed on your computer. If you already use it, start it in the folder where you will work. Ask it to check your R version and install lisaR from GitHub with the commands above.
In the ChatGPT desktop app, select Work and keep Work locally selected for an installation on your computer. Give it the repository address and ask it to check that it can use your local R installation. When those tools are available, it can help carry out the same commands. Otherwise, use its explanation and run the commands yourself in RStudio. Installing a package in a cloud session does not install it in your local R library.
Claude Code can read project files and run commands. If you already use it, open the folder containing the project and ask it to check R and follow the GitHub installation steps above.
If you use a Claude chat for scientific advice, ask it to explain these R commands and help interpret an installation error. Run the commands in your own R session unless the chat has access to that installation and can execute them there.
For any of these tools, a suitable request is:
Install lisaR from https://github.com/DBM-OlmedaLab/lisaR on this computer. First check that R is version 4.1 or later and show which local R library will be used. Follow the dependency and GitHub commands in this guide, then load lisaR and show its version and library path. Check that the included dictionaries and sample project are present. Do not perform a second source-file installation.
The installation above does not rebuild HTML vignettes. Read the guides online, where each guide also has a PDF download. You can open the Quick Start directly from R:
The source files are in the repository’s vignettes/
directory, in .Rmd format. vignette() lists
compiled HTML only when the package was built with vignettes. You do not
need those HTML files or another installation to run the examples.
| Message or problem | Action |
|---|---|
| A package dependency is missing | Run the dependency step in the same R library, then install lisaR again. |
| R and Bioconductor versions are incompatible | Select the Bioconductor release supported by the active R version. |
| A source dependency requires compilation | Install the platform’s R build tools: Rtools on Windows, the command-line tools on macOS, or the appropriate compiler packages on Linux. |
| Pandoc is missing during a vignette build | Use the Pandoc supplied by RStudio or install it before rebuilding documentation. Pandoc is not needed for an ordinary lisaR analysis. |
| An older lisaR version still appears | Restart R and check the library returned by
find.package("lisaR"). |
See Troubleshooting for configuration and execution errors after installation.