skill

IOBRskill

A skill that enables AI to better invoke IOBR for tumor microenvironment analysis.

IOBR/IOBRskillcommunitylistedunreviewed

Set up

Skills are installed by placing their folder where your agent loads skills.

Download ZIP from GitHub Open the repository

The download comes from github.com/IOBR/IOBRskill itself, not from our server, so what you get is exactly what its author published.

Install instructions come from the README

Skills do not use an MCP config. Follow the steps in the README below.

README

synced Oct 22, 2026

IOBRskill

A Claude Code skill for end-to-end tumor microenvironment (TME) analysis of bulk transcriptome data using the IOBR R package.

IOBRskill automates the complete TME analysis pipeline — from data preprocessing and annotation through immune cell deconvolution, signature scoring, statistical analysis, and publication-quality visualization — all through natural language interaction with Claude Code.

What It Does

IOBRskill guides Claude Code through a standardized 6-phase TME analysis workflow:

Phase	Step	Description
0	Plan	Generate `IOBR-pipeline.md` — ASCII tree of all scripts, inputs/outputs, and expected results
1	`01-data_preprocessing.R`	Data loading, normalization, probe annotation, QC, batch correction, sample matching, pdata summary
2	`02-tme_deconvolution.R`	Immune cell deconvolution (8 methods: CIBERSORT, CIBERSORT-ABS, MCPcounter, EPIC, xCell, ESTIMATE, TIMER, quanTIseq, IPS)
3	`03-signature_analysis.R`	Gene signature scoring (ssGSEA/PCA/Z-score), Hallmark/KEGG/GO/Reactome pathway scoring, TME subtype clustering
4	`04-statistical_analysis.R`	Merge → scale → cluster → Wilcoxon/Kruskal tests → Cox survival → correlation matrix. All results saved to `04-figs/data/`
5	`05-visualization.R`	Nature-style publication figures: Fig01–Fig09 (barplot, heatmap, forest, correlation, boxplot, KM) — PNG 300dpi + PDF
6	Wrap-Up Note	Generate `05-note/IOBR-analysis-README.md` with actual output tree, figure count, and key findings summary

Decision Points

IOBRskill pauses at key steps to let the user choose:

Phase 1: Data type (Count/TPM/Array) and species (human/mouse)
Phase 2: Deconvolution method(s) — CIBERSORT + MCPcounter + ESTIMATE recommended
Phase 3: Scoring method (ssGSEA default) and signature collection
Phase 5: Color palette preference

Requirements

Claude Code CLI (latest version recommended)
R >= 4.2.0
IOBR R package >= 2.2.0

Installation

Step 1 -- Install the Skill

Choose one of the following methods:

This makes IOBRskill available in all your Claude Code sessions:

git clone https://github.com/IOBR/IOBRskill.git ~/.claude/skills/IOBRskill

That's it. Claude Code automatically discovers skills under ~/.claude/skills/.

Install into a specific project so the skill only activates in that workspace:

cd /path/to/your/project
git clone https://github.com/IOBR/IOBRskill.git

Or add it as a git submodule:

cd /path/to/your/project
git submodule add https://github.com/IOBR/IOBRskill.git IOBRskill

Download the repository and copy the IOBRskill/ folder to your preferred location:

# Download
wget https://github.com/IOBR/IOBRskill/archive/refs/heads/main.zip
unzip main.zip

# Copy to global skills
cp -r IOBRskill-main/ ~/.claude/skills/IOBRskill

# Or copy to a specific project
cp -r IOBRskill-main/ /path/to/your/project/IOBRskill

Step 2 -- Install IOBR R Package

IOBRskill depends on the IOBR R package. Install it in your R environment:

# Install BiocManager if needed
if (!requireNamespace("BiocManager", quietly = TRUE))
    install.packages("BiocManager")

# Install IOBR from GitHub (latest version)
BiocManager::install("IOBR/IOBR")

# Or install from CRAN (stable release)
install.packages("IOBR")

# Verify
packageVersion("IOBR")
# Should print: [1] '2.2.0' or higher

If you plan to use CIBERSORT deconvolution, also install:

BiocManager::install("preprocessCore")

Step 3 -- Verify

Open Claude Code and test the skill:

/IOBRskill

You should see the IOBR analysis pipeline launched. If you get an IOBR not found error, run:

Rscript -e 'library(IOBR); cat("IOBR", as.character(packageVersion("IOBR")), "OK\n")'

Trigger Conditions

IOBRskill activates when you:

Type /IOBRskill directly
Mention any of these keywords in your prompt:
- English: tumor microenvironment analysis, TME deconvolution, immune infiltration, ligand-receptor analysis, CIBERSORT, MCPcounter, ESTIMATE, immune cell deconvolution
- Chinese: 肿瘤微环境分析, 肿瘤微环境解析, 免疫浸润分析, 受体配体分析, 通路分析

How to Use

Simply describe what you want to analyze. IOBRskill will interactively guide you through the pipeline:

Example prompts

# Start a full TME analysis
/IOBRskill

# With your own data
"I have a raw count matrix at ~/data/expr.csv from 50 human tumor samples.
I want to do a comprehensive TME analysis."

# Specific method
"Run CIBERSORT and MCPcounter deconvolution on my expression matrix"

# Focus on specific analysis
"I want to analyze ligand-receptor interactions in my TME data"

Output Structure

IOBRskill creates a standardized directory structure. Here is an example from IMvigor210 (bladder cancer immunotherapy cohort, 348 samples, RNA-seq):

IMvigor210/
├── 01-script/
│   ├── 01-data_preprocessing.R       # Count → TPM, gene annotation, log2, QC
│   ├── 02-tme_deconvolution.R        # 9-method deconvolution + merge
│   ├── 03-signature_analysis.R       # ssGSEA: TME signatures + Hallmark + KEGG
│   ├── 04-statistical_analysis.R     # Merge, scale, cluster, Wilcoxon, survival
│   └── 05-visualization.R           # Fig01–Fig09 (all figures)
├── 02-input/
│   ├── annotated_eset.csv            # Log2-normalized expression (genes x samples)
│   ├── pdata.csv                     # Sample phenotype data
│   └── pdata_summary.csv             # Variable type/level summary
├── 03-tme/
│   ├── tme_cibersort.csv             # 22 immune cell fractions per sample
│   ├── tme_mcpcounter.csv            # 8 cell population scores per sample
│   ├── tme_estimate.csv              # Stromal/Immune/Tumor purity scores
│   ├── tme_epic.csv                  # 8 cell types incl. cancer fraction
│   ├── tme_xcell.csv                 # 64 cell type scores
│   ├── tme_quantiseq.csv             # 10 cell type fractions
│   ├── tme_ips.csv                   # 4 immunotherapy response axes
│   ├── sig_score_tme.csv             # TME signature scores (ssGSEA)
│   ├── sig_score_pathway.csv         # Hallmark + GO + KEGG + Reactome scores
│   └── tme_sig_combine.csv           # All results merged by sample ID
├── 04-figs/
│   ├── Fig01-barplot_cibersort.png/pdf        # CIBERSORT cell composition barplot
│   ├── Fig02-barplot_epic.png/pdf             # EPIC cell composition barplot
│   ├── Fig03a-heatmap_cibersort.png/pdf       # CIBERSORT heatmap (z-score)
│   ├── Fig03b-heatmap_mcpcounter.png/pdf      # MCPcounter heatmap (z-score)
│   ├── Fig03c-heatmap_xcell.png/pdf           # xCell heatmap (z-score)
│   ├── Fig04a-forest_cibersort.png/pdf        # CIBERSORT survival forest (if survival)
│   ├── Fig04b-forest_other_tme.png/pdf        # MCPcounter+EPIC+xCell forest (if survival)
│   ├── Fig04c-forest_tme_signature.png/pdf    # TME signatures forest (if survival)
│   ├── Fig04d-forest_go_kegg.png/pdf          # GO/KEGG/Hallmark forest (if survival)
│   ├── Fig05-cor_matrix.png/pdf               # TME cell correlation matrix heatmap
│   ├── Fig06a-heatmap_cibersort_subtype.png/pdf  # CIBERSORT heatmap by TME subtype
│   ├── Fig06b-heatmap_mcpcounter_subtype.png/pdf # MCPcounter heatmap by TME subtype
│   ├── Fig06c-heatmap_xcell_subtype.png/pdf   # xCell heatmap by TME subtype
│   ├── Fig06d-km_subtype.png/pdf              # KM plot by TME subtype (if survival)
│   ├── Fig07a-boxplot_cibersort_subtype.png/pdf   # CIBERSORT subtype boxplots
│   ├── Fig07b-boxplot_mcpcounter_subtype.png/pdf  # MCPcounter subtype boxplots
│   ├── Fig07c-boxplot_xcell_subtype.png/pdf   # xCell subtype boxplots
│   ├── Fig07d-boxplot_tme_signature_subtype.png/pdf # TME signatures subtype boxplots
│   ├── Fig07e-boxplot_pathway_subtype.png/pdf  # GO/KEGG/Hallmark subtype boxplots
│   ├── Fig08a-top10_<variable>.png/pdf        # Top 10 differential boxplots (if categorical)
│   ├── Fig08b-top10_<variable>.png/pdf        # Top 10 differential boxplots (if categorical)
│   ├── kmplot/                                # KM plots (most/least significant variables)
│   └── data/                                  # Statistical result tables
│       ├── 01-tme_pdata_merged.csv
│       ├── 02-tme_scaled.csv
│       ├── 03-tme_subtype.csv
│       ├── 04-diff_<variable>.csv
│       ├── 05-surv_cibersort.csv
│       ├── 05-surv_other_tme.csv
│       ├── 05-surv_tme_signatures.csv
│       ├── 05-surv_pathways.csv
│       └── 06-cor_matrix.csv
├── 05-note/
│   ├── IOBR-pipeline.md               # Analysis plan (generated in Phase 0)
│   └── IOBR-analysis-README.md        # Output tree + key findings summary (Phase 6)
└── 06-log/
    ├── 01-data_preprocessing.log      # Execution logs with IOBR citation header
    ├── 02-tme_deconvolution.log
    ├── 03-signature_analysis.log
    ├── 04-statistical_analysis.log
    └── 05-visualization.log

Figure Logic

Condition	Figure	Content
Always	Fig01–02	CIBERSORT + EPIC fraction barplots
Always	Fig03a–c	Heatmaps without subtype (CIBERSORT, MCPcounter, xCell)
IF survival data	Fig04a–d	Forest plots (CIBERSORT, other TME, signatures, pathways)
Always	Fig05	TME cell correlation matrix
Always	Fig06a–c	Heatmaps by TME subtype
IF survival data	Fig06d	KM plot by TME subtype
Always	Fig07a–e	Most positive cell per subtype boxplots
IF categorical vars	Fig08a–b...	Top 10 differential boxplots (Wilcoxon/Kruskal)
IF survival data	Fig09/kmplot/	8 KM plots (extreme HR from each group)

Key Rules

sig_forest: Must filter abs(HR) > 10 before passing data -- extreme HR causes blank plots
ALL boxplots: Use sig_box() with fixed params: palette = "paired1", show_pvalue = TRUE, angle_x_text = 60, hjust = 1, size_of_pvalue = 3, size_of_font = 5
Heatmaps: Variables as rows, samples as columns, z-score scaled, capped at +/-2
Figures: Always dual-format PNG (300dpi) + PDF
Citation header: Every R script and log file must include the IOBR citation

Supported Methods

TME Deconvolution

Method	Cell Types	Description
CIBERSORT	22	Gold standard immune profiling (LM22)
CIBERSORT-ABS	22	Absolute mode (fraction-independent)
MCPcounter	8	Stromal + immune quantification
EPIC	8	Includes cancer cell fraction
xCell	64	Broadest cell type coverage
ESTIMATE	4 scores	Stromal/Immune/Tumor purity
TIMER	6	Cancer type-specific
quanTIseq	10	M1/M2 macrophage resolution
IPS	4 axes	Immunotherapy response prediction

Signature Scoring

Method	Description
ssGSEA	Single-sample GSEA (recommended, default)
PCA	Principal component analysis
Z-score	Z-score normalization
Integration	Combined method

Built-in Gene Signatures

IOBR ships with 323 curated gene signatures organized into 35+ categories:

TME signatures (186): immune checkpoint, T cell exhaustion, EMT, TME scores
Hallmark pathways (50)
Metabolism pathways (113)
GO Biological Process (7,658), GO Cellular Component, GO Molecular Function
KEGG pathways (186)
Reactome pathways (1,615)
Cell-type-specific signatures from CIBERSORT, MCPcounter, EPIC, xCell, quanTIseq
Published signatures: Rooney et al, Bindea et al, Li et al, Peng et al, and more

Dependencies

Package	Required	Install
IOBR >= 2.2.0	Yes	`BiocManager::install("IOBR/IOBR")`
R >= 4.2.0	Yes	--
preprocessCore	For CIBERSORT	`BiocManager::install("preprocessCore")`
limSolve	For quanTIseq	`install.packages("limSolve")`
NbClust	For tme_cluster()	`install.packages("NbClust")`
FactoMineR	For iobr_pca()	`install.packages("FactoMineR")`
ggplot2	For visualization	`install.packages("ggplot2")`
pheatmap	For heatmaps	`install.packages("pheatmap")`
patchwork	For multi-panel figures	`install.packages("patchwork")`
survminer	For KM plots	`install.packages("survminer")`

Skill Structure

IOBRskill/
├── SKILL.md                               # Main skill instructions (full pipeline)
├── README.md                              # This file
├── evals/
│   └── evals.json                         # Test cases
├── references/
│   ├── functions.md                       # IOBR function parameter reference
│   ├── palettes.md                        # Color palette selection guide
│   ├── iobr_built_in_data.md              # Built-in data & signature catalog
│   └── iobr_pipeline_template.R           # Full production pipeline template
└── scripts/
    └── IOBRskill.png                      # Pipeline overview diagram

Citation

If you use IOBRskill in your research, please cite IOBR:

Zeng DQ, Fang YR, ... , Yu GC, Liao WJ. Enhancing Immuno-Oncology Investigations Through Multidimensional Decoding of Tumour Microenvironment with IOBR 2.0. Cell Reports Methods, 2024. https://doi.org/10.1016/j.crmeth.2024.100910

Fang YR, ..., Liao WJ, Zeng DQ. Systematic Investigation of Tumor Microenvironment and Antitumor Immunity With IOBR. Med Research, 2025. https://doi.org/10.1002/mdr2.70001

Contact

Dongqiang Zeng -- interlaken@smu.edu.cn
Issues: https://github.com/IOBR/IOBRskill/issues

IOBRskill

Set up

README

IOBRskill

What It Does

Decision Points

Requirements

Installation

Step 1 -- Install the Skill

Step 2 -- Install IOBR R Package

Step 3 -- Verify

Trigger Conditions

How to Use

Example prompts

Output Structure

Figure Logic

Key Rules

Supported Methods

TME Deconvolution

Signature Scoring

Built-in Gene Signatures

Dependencies

Skill Structure

Links

Citation

Contact