RStudio Server

Two variants are available depending on how R is installed:

Variant	R Installation	Best For
`rstudio-server`	Posit R image overlays	Normal `install.packages()`, GitHub, Bioconductor from source
`rstudio-server-conda`	Conda (`mm-install r-base`)	All packages via conda-forge/bioconda, reproducible env export

condatainer helper -u
condatainer helper rstudio-server -w         # Posit R
condatainer helper rstudio-server-conda -w   # Conda R

See Helper Scripts for SSH port forwarding setup, resource flags, configuration, and reuse mode.

rstudio-server (Posit R)

Uses Posit R image overlays — a separate r<version> overlay is auto-installed alongside rstudio-server and build-essential.

Writable Overlay

# Create a 30G writable overlay in the current directory
condatainer o -s 30g

# With Python (needed for reticulate):
condatainer o -s 30g -- python=3.11 conda

Flags

condatainer helper rstudio-server --help

Flag	Description	Default
`--rversion,-r`	R version (e.g. `4.5.2`, `4.4`)	latest available
`--password,-p`	RStudio password (empty = no auth)	(empty)

Partial versions are accepted: -r 4.4 uses the latest 4.4.x available. See Helper Scripts for resource and overlay flags.

Installing R Packages

The build-essential overlay provides common system libraries. Install packages normally:

install.packages("tidyverse")
BiocManager::install("DESeq2")
pak::pak("user/repo")

Rprofile Setup

When no *.Rproj file exists in the working directory, CondaTainer creates:

A *.Rproj file with default settings
A .Rprofile configured for Posit Public Package Manager (P3M) binary packages

If *.Rproj already exists, .Rprofile is not modified. To use only CRAN:

.set_repository_options(repo = "cran", latest_cran = TRUE)

Missing System Libraries

Create a custom overlay for extra system libraries — see Custom OS Overlays:

condatainer helper rstudio-server -o additional-deps.sqf

Run R Without RStudio

condatainer exec -o r4.4.3 -o build-essential -o env.img Rscript script.R
condatainer exec -o r4.4.3 -o build-essential -o r-deps.sqf -o env.img Rscript script.R

rstudio-server-conda (Conda R)

R is installed in the writable overlay via mm-install r-base. No separate R overlay is needed.

Note

Always use mm-install (Conda) to install R packages in this variant. Do not use install.packages() — it may conflict with the Conda-managed environment.

Writable Overlay with R

# Create a 30G overlay with R pre-installed
condatainer o -s 30g -- r-base=4.4 r-tidyverse

# Enter the overlay to install more packages
condatainer e

Inside the overlay:

mm-pin r-base   # pin R version to prevent accidental updates
mm-install r-seurat r-patchwork bioconductor-clusterprofiler
mm-install python=3.11 conda   # add Python for reticulate

Package Management

Conda naming conventions:

CRAN: r-<name> (e.g. r-ggplot2, r-dplyr)
Bioconductor: bioconductor-<name> (e.g. bioconductor-deseq2)

mm-search r-presto
mm-install r-ggplot2 bioconductor-deseq2
mm-export --no-builds > conda-env.yml   # export full environment

Run R Without RStudio

condatainer exec -o env.img Rscript script.R

Once running, open the project in R:

rstudioapi::openProject("/scratch/user/my-project")

Common Issues

UID conflict (UID = 1000)

The Posit R container image reserves UID 1000 for its internal user. If your account UID is 1000, rstudio-server will not start — use rstudio-server-conda instead.