rnaseq-plot

RNA-seq downstream plotting for MiniMax Code (formerly published as rgraph). Parameterized R scripts (ggplot2, pheatmap, clusterProfiler, edgeR, limma, WGCNA, …) are exposed as MCP tools and rendered by the user's local Rscript to png + pdf.

The Plugin consumes the user's own count/FPKM tables. It does not ship genomes or experimental matrices. tests/data/ is a tiny synthetic g1–g10 table for smoke tests.

MCP tool names remain rgraph_* (for example rgraph_volcano) so existing prompts keep working.

Try it

I have gene_count.csv and sample_group.csv. Run DESeq2 for treatment vs control, then draw a
volcano plot (label top 8 genes, padj < 0.05, |log2FC| > 1) and a clustered heatmap of DEGs.

Expected result: the agent calls rgraph_env, then rgraph_diff(method="deseq2") and rgraph_volcano / rgraph_heatmap. Output png+pdf paths are returned. DEG calls use padj by default. If R or a package is missing, the tool returns install commands instead of crashing.

Requirements

• Python 3.10+ and uv on PATH.
• R with Rscript on PATH, or set RGRAPH_RSCRIPT to the Rscript executable.
• Common R packages: ggplot2, pheatmap, edgeR or DESeq2 or limma, clusterProfiler as needed. Missing packages are reported with CRAN/Bioconductor install lines.
• Windows, macOS, and Linux.

Data and network

• Default analyses are local: user CSVs in, png/pdf out. No telemetry.
• rgraph_ppi may contact STRING (string-db.org) when the user asks for a PPI edge table.
• No credentials in the package.

License

MIT. See LICENSE. Source: https://github.com/Presisitence/rnaseq-plot-mcp