tooluniverse-stem-cell-organoid — Claude Code skill: what it does
tooluniverse-stem-cell-organoid guides stem cell protocols, pluripotency verification, and organoid fidelity scoring using ToolUniverse biomedical tools.
Apache-2.0 226 installs ★ 1,798 updated 2026-10-24
npx skills add mims-harvard/ToolUniverse --skill tooluniverse-stem-cell-organoid -a claude-codeWhat tooluniverse-stem-cell-organoid does
This skill configures a structured workflow for stem cell biology and directed differentiation. It directs the model to interrogate tools like CellMarker_search_by_cell_type and kegg_search_pathway rather than relying on internal memory. Protocols cover pathways such as WNT, BMP/TGF-beta, Notch, and FGF with specific modulators like CHIR99021 or Noggin.
Without cellxgene-census installed via pip, the CELLxGENE Census tools cannot pull single-cell expression matrices. The reasoning framework evaluates organoid models against adult versus fetal maturation markers and grades findings across formal evidence tiers (T1 through T4).
What it helps with
- retrieve validated lineage markers using CellMarker queries
- sequence pathway modulators for directed stem cell differentiation
- evaluate organoid structural and functional fidelity against real tissue
- locate relevant single-cell RNA-seq datasets in GEO and Human Cell Atlas
Example prompts for the agent
- Query CellMarker for pancreatic beta cell markers and outline the required signaling cascade
- Assess cortical organoid fidelity based on cellular diversity and adult maturation markers
- Find human iPSC single-cell RNA-seq datasets using GEO and CELLxGENE tools
How to install tooluniverse-stem-cell-organoid
1Skills CLI (Vercel)
npx skills add mims-harvard/ToolUniverse --skill tooluniverse-stem-cell-organoid -a claude-code2Manually: put the files in place
~/.claude/skills/tooluniverse-stem-cell-organoid/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/tooluniverse-stem-cell-organoid/
1Skills CLI (Vercel)
npx skills add mims-harvard/ToolUniverse --skill tooluniverse-stem-cell-organoid -a codex2Manually: put the files in place
.agents/skills/tooluniverse-stem-cell-organoid/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add mims-harvard/ToolUniverse --skill tooluniverse-stem-cell-organoid -a cursor2Manually: put the files in place
.cursor/skills/tooluniverse-stem-cell-organoid/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install mims-harvard/ToolUniverse skills/tooluniverse-stem-cell-organoid2Manually: put the files in place
.github/skills/tooluniverse-stem-cell-organoid/SKILL.md1Skills CLI (Vercel)
npx skills add mims-harvard/ToolUniverse --skill tooluniverse-stem-cell-organoid -a gemini-cli2Platform CLI
gemini skills install https://github.com/mims-harvard/ToolUniverse --path skills/tooluniverse-stem-cell-organoidManually: put the files in place
.devin/skills/tooluniverse-stem-cell-organoid/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add mims-harvard/ToolUniverse --skill tooluniverse-stem-cell-organoid -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about tooluniverse-stem-cell-organoid
How to install tooluniverse-stem-cell-organoid in Claude Code?
Run npx skills add mims-harvard/ToolUniverse --skill tooluniverse-stem-cell-organoid -a claude-code. The definition will be placed in ~/.claude/skills/tooluniverse-stem-cell-organoid/SKILL.md.
What is the purpose of this skill?
It gives coding agents standard operating procedures to investigate stem cell markers, differentiation pathways, and organoid validation through biological database tools.
Are there extra packages required for full functionality?
Yes, using the CELLxGENE_get_census_versions tool requires the cellxgene-census Python package installed via pip install cellxgene-census.
Source text
SKILL.md · Apache-2.0
---
name: tooluniverse-stem-cell-organoid
description: Stem cell, iPSC, and organoid research — pluripotency markers, differentiation protocol pathways, lineage commitment factors, organoid model selection. Use for iPSC characterization, differentiation protocol design via developmental-pathway recapitulation, and organoid-model selection for disease modeling.
disable-model-invocation: true
---
# Stem Cell & Organoid Research
Pipeline for investigating stem cell biology, iPSC characterization, organoid models, and cell differentiation using ToolUniverse tools.
## Reasoning Strategy
Stem cell differentiation follows developmental biology — to make any target cell type from iPSCs, the protocol must mimic the embryonic signaling pathway that generates that cell type in vivo. For neural induction: inhibit BMP and TGF-beta (dual SMAD inhibition). For cardiomyocytes: activate WNT then inhibit WNT. For pancreatic beta cells: activate Activin/Nodal → FGF → Notch inhibition → BMP in sequence. The order and timing of growth factors matters critically — adding BMP4 during neural induction will redirect cells toward mesoderm. Mouse and human stem cells differ in their signaling requirements (LIF/STAT3 for mouse naive pluripotency; FGF/ERK for human primed pluripotency), so protocols are not interchangeable. Organoids recapitulate some but not all organ features — always assess maturation state (fetal vs. adult gene expression) before drawing disease-relevance conclusions.
**LOOK UP DON'T GUESS**: Do not assume which markers define a target cell type or which signaling pathway drives differentiation — query `CellMarker_search_by_cell_type` for markers and `kegg_search_pathway` for the relevant pathway. Do not assume organoid fidelity; look up published CellxGene or HCA atlas data for comparison.
**Key principles**:
1. **Marker-based identity** — stem cell identity is defined by marker expression profiles (OCT4, SOX2, NANOG for pluripotency)
2. **Differentiation is a trajectory** — not a binary state; analyze intermediate progenitor stages
3. **Organoid ≠ organ** — organoids recapitulate some but not all organ features; always note limitations
4. **Species matters** — mouse and human stem cells differ in signaling requirements
5. **Evidence grading** — T1: validated in clinical iPSC study, T2: functional assay (teratoma, engraftment), T3: marker expression only, T4: computational prediction
---
## Core Tools
| Tool | Use For |
|------|---------|
| `CELLxGENE_get_census_versions` | Discover CELLxGENE Census release versions; then use `CELLxGENE_get_cell_metadata` / `CELLxGENE_get_expression_data` for specific cells / genes. **Requires `cellxgene-census` package (`pip install cellxgene-census`). May not be installed by default.** |
| `CellMarker_search_by_cell_type` | Cell type marker genes. **Requires `operation="search_by_cell_type"`, `cell_name=` (NOT `cell_type=`)** |
| `CellMarker_search_by_gene` | Which cell types express a gene. **Requires `operation="search_by_gene"`, `gene_symbol=`** |
| `hca_search_projects` | Human Cell Atlas organoid/development projects |
| `GEO_search_rnaseq_datasets` | Find stem cell RNA-seq datasets |
| `kegg_search_pathway` | Differentiation signaling pathways (WNT, Notch, Hedgehog) |
| `ReactomeAnalysis_pathway_enrichment` | Pathway analysis of stem cell gene sets |
| `STRING_get_network` | Pluripotency/differentiation gene networks |
| `OpenTargets_get_associated_targets_by_disease_efoId` | Disease genes for organoid disease modeling |
| `PubMed_search_articles` | Stem cell and organoid literature |
| `search_clinical_trials` | iPSC-based clinical trials |
---
## Workflow
```
Phase 0: Define the Question
Pluripotency? Differentiation? Disease modeling? Drug screening?
|
Phase 1: Cell Identity & Markers
CellMarker → pluripotency/lineage markers → verify identity
|
Phase 2: Differentiation Pathways
KEGG/Reactome → WNT, Notch, BMP, FGF signaling
|
Phase 3: Atlas & Dataset Discovery
CellxGene/HCA → reference datasets for target cell type
|
Phase 4: Disease Modeling (if applicable)
OpenTargets → disease genes → organoid recapitulation assessment
|
Phase 5: Report
Evidence-graded findings with clinical translation potential
```
### Phase 1: Cell Identity & Markers
**Pluripotency markers** (must be co-expressed): OCT4 (POU5F1), SOX2, NANOG (essential); SSEA-4, TRA-1-60 (human surface markers). KLF4 and MYC are Yamanaka factors but also expressed in somatic cells — do not rely on them alone. Use `CellMarker_search_by_cell_type` to retrieve the full validated marker set for any target cell type.
**Lineage markers**: Ectoderm → PAX6/SOX1 (early), MAP2/TUBB3 (neurons); Mesoderm → TBXT/MIXL1 (early), CD34 (blood); Endoderm → SOX17/FOXA2 (early), PDX1/NKX6.1 (pancreas). Retrieve current marker lists from CellMarker rather than relying on memory.
### Phase 2: Differentiation Pathways
Key signaling pathways for directed differentiation:
| Pathway | KEGG ID | Role in Stem Cells | Common Modulators |
|---------|---------|-------------------|-------------------|
| WNT signaling | hsa04310 | Pluripotency maintenance (canonical) vs differentiation (non-canonical) | CHIR99021 (activator), IWP-2 (inhibitor) |
| Notch signaling | hsa04330 | Lateral inhibition, fate decisions | DAPT (gamma-secretase inhibitor) |
| BMP/TGF-beta | hsa04350 | Mesoderm/trophectoderm induction | BMP4 (activator), Noggin (inhibitor) |
| FGF signaling | hsa04010 | Self-renewal, neural induction | bFGF (activator), SU5402 (inhibitor) |
| Hedgehog | hsa04340 | Patterning, organoid maturation | SAG (activator), cyclopamine (inhibitor) |
| Hippo/YAP | hsa04390 | Mechanotransduction, organoid size | Verteporfin (YAP inhibitor) |
### Phase 3: Atlas & Dataset Discovery
```python
# Find stem cell single-cell datasets
CELLxGENE_get_census_versions() # discover available Census releases, then use CELLxGENE_get_cell_metadata / CELLxGENE_get_expression_data
hca_search_projects(query="organoid")
GEO_search_rnaseq_datasets(query="iPSC differentiation neural", organism="Homo sapiens")
```
### Phase 4: Organoid Model Assessment
**Organoid fidelity scoring** — how well does the organoid recapitulate the organ?
| Feature | High Fidelity (3) | Moderate (2) | Low (1) |
|---------|------------------|-------------|---------|
| Cell type diversity | All major cell types present | Most cell types, missing rare ones | Only 1-2 cell types |
| Architecture | Self-organized, correct spatial arrangement | Partial organization | Disorganized aggregate |
| Function | Measurable organ function (secretion, contraction, electrophysiology) | Some functional markers | Marker expression only |
| Maturation | Adult-like gene expression profile | Fetal-like | ESC-like (failed differentiation) |
| Disease relevance | Recapitulates patient phenotype | Some disease features | No disease phenotype |
---
## Evidence Grading
| Grade | Criteria | Example |
|-------|---------|---------|
| **T1** | Clinical iPSC study or approved therapy | iPSC-derived RPE for macular degeneration (Mandai 2017) |
| **T2** | Functional validation (teratoma, engraftment, drug response) | Organoid drug screening with patient-specific response |
| **T3** | Marker expression + morphology | iPSC colony expressing OCT4/SOX2/NANOG |
| **T4** | Computational prediction or single-marker evidence | Predicted pluripotent by gene expression classifier |
### Synthesis Questions
1. **Is the cell identity verified?** (co-expression of 3+ pluripotency markers, or lineage-appropriate markers)
2. **Is the differentiation protocol reproducible?** (published, peer-reviewed, with quantified efficiency)
3. **Does the organoid model the disease?** (patient-derived iPSC shows disease phenotype in organoid)
4. **What are the translational barriers?** (scalability, maturation, immune compatibility, tumorigenicity)
5. **What's the best reference dataset?** (CellxGene atlas for comparison)
---
## Limitations
- **No organoid protocol database** — protocols are scattered across publications; use PubMed search
- **Maturation gap** — most organoids resemble fetal, not adult tissue; always note maturation state
- **Batch variability** — iPSC-derived cells vary between passages and donor lines
- **No direct culture tools** — this skill analyzes published data and designs experiments; it does not control bioreactors
- **Species differences** — mouse ESCs require LIF; human ESCs require bFGF. Don't mix protocols
© mims-harvard · Apache-2.0 · mims-harvard/ToolUniverse