RStudio analysis
RStudio is the platform’s independent, optional R execution stack: a browser-based R analysis environment for R power users and for statistics and plotting that benefits from interactive exploration. It shares the same workspace files, tasks, and artifacts as the manual and Agent paths (see Usage paths) and can be enabled as needed.
The self-hosted R stack
The platform maintains three versioned R containers, each with a distinct role:
| Runtime | Version | Positioning | Best for |
|---|---|---|---|
| Main runtime | R 4.6.1 / Bioc 3.23 | The mainline RStudio Server; the platform’s web entry (/rstudio/) points here | General R analysis: statistical modeling, ggplot2 plotting, data wrangling ships common stacks such as enrichment, DESeq2/edgeR, Seurat, and CellChat plus compatible packages like CMplot, fgsea, and WGCNA |
| Bridge container | R 4.5.3 / Bioc 3.22 | Compatibility bridge line | Packages depending on older Bioconductor releases (Signac / Seurat / chromVAR and other scATAC stacks) for cases where they don’t work or behave differently under 4.6.1 |
| ArchR container | R 4.4.1 / Bioc 3.19 | ArchR-dedicated line (ArchR 1.0.3) | Single-cell ATAC analysis that follows the ArchR pipeline |
How to choose: use the main runtime by default; switch to the bridge only when a package explicitly requires older Bioconductor dependencies; use the ArchR container for ArchR pipelines.
[Screenshot coming: runtime selection when creating an RStudio session]
Creating and using sessions
- Enter the RStudio entry from the platform and pick the runtime you need (the main runtime by default);
- Log in with your platform account and the familiar RStudio Server layout opens in your browser (Console / Script / Environment / Files);
- Workspace files and task artifacts are visible inside the session and can be read directly; artifacts you save there can be referenced by later platform tasks;
- For long jobs, write a script and run it with logs preserved rather than executing everything interactively in the Console.
[Screenshot coming: RStudio Server session main interface]
What Agent-controlled R script execution looks like
The Agent doesn’t turn RStudio into a chat window — instead it can help you maintain and run R analyses under controlled conditions:
- Task-based execution: R analyses are submitted and run as platform tasks, not run bare inside the conversation;
- Logs you can inspect: task stdout/stderr is fully recorded, and the Agent can read the logs to help pinpoint errors (a missing package, a wrong path, etc.);
- Artifact retrieval: outputs return to the platform’s artifact list, ready to preview, download, or feed into the next step;
- Strict confirmation: when the Agent updates or writes R scripts, that is an workspace file operation and follows workspace boundaries and session scope (see MCP governance);
- No guessing about packages: the Agent cannot probe the R environment directly; when package versions or availability matter, it probes through a platform task or asks you to run
packageVersion()in the session.
Typical loop: ask the Agent to draft a plotting or statistics script → after your confirmation it runs it as a platform task → the Agent reads the logs and artifacts, discusses parameter adjustments with you, and iterates the next round.
Who it’s for
- R power users: you already have mature R workflows and want platform data processed directly in R;
- Interactive statistics and plotting: exploratory data analysis with live iteration, or publication-grade ggplot polishing;
- Packages beyond the platform: R capabilities the tool catalog doesn’t cover can be used in the session (the bridge / ArchR lines solve version compatibility);
- R users who collaborate with the Agent: hand repetitive script maintenance, execution, and log triage to the Agent while you focus on statistics and biology.
Common questions
Which runtime should I use?
When in doubt, use the main runtime (r4.6.1). If an error says a package is missing and that package is only published for older Bioconductor, switch to the bridge (r4.5.3); for ArchR pipelines use the ArchR container (r4.4.1).
How do artifacts from RStudio get back into the platform?
Files saved in the session working directory can be referenced as task inputs; for formal analyses it’s best to run through platform tasks so artifacts land in the artifact list automatically.
Can the Agent install R packages for me?
No. The Agent can only execute authorized R analyses and script maintenance under control; when a package is missing it will tell you the package name and version requirements, and you install it in the session or fall back to platform tools.
Next steps
- Manual mode: form-driven analysis in the tool catalog.
- Agent-assisted mode: how the rstudio-analysis-runtime skill triggers.
- Native workflows: for standard pipelines, prefer direct submission via native workflows.