tooluniverse-lipidomics — Claude Code skill
tooluniverse-lipidomics is a Claude Code skill that identifies and classifies lipids, maps metabolic pathways, and retrieves disease associations for lipidomics analysis.
Apache-2.0 238 installs ★ 1,798 updated 2026-10-24
npx skills add mims-harvard/ToolUniverse --skill tooluniverse-lipidomics -a claude-codeWhat tooluniverse-lipidomics does
tooluniverse-lipidomics connects multiple lipid and disease databases to support end-to-end lipidomics analysis. Given a lipid name, mass, or abbreviation, the skill retrieves its structure and assigns it to one of eight LIPID MAPS categories: fatty acyls, glycerolipids, glycerophospholipids, sphingolipids, sterol lipids, prenol lipids, saccharolipids, and polyketides. Each category has distinct metabolic roles and disease contexts. The skill then maps the lipid to biosynthesis and degradation enzymes via KEGG and Reactome, and pulls disease associations from CTD and DisGeNET.
Core tools include LipidMaps_search_by_name, LipidMaps_get_compound_by_id, SwissLipids_search, HMDB_search, kegg_search_pathway, CTD_get_gene_diseases, and PubMed_search_articles. For specialized lipid classes—sphingolipids, eicosanoids, steroids, fatty acids—the skill already knows the relevant pathways. The skill queries databases rather than assuming lipid IDs or disease links.
Requires network access to LIPID MAPS, HMDB, KEGG, Reactome, PubChem, and CTD. Not for general metabolomics (amino acids, sugars, organic acids); use a separate metabolomics skill for those.
What it helps with
- Identify a lipid species from mass spectrometry m/z and retention time
- Find metabolic pathways that involve ceramide or sphingomyelin biosynthesis and degradation
- Retrieve lipid biomarkers for Alzheimer's disease and cardiovascular disease from HMDB
- Map genes and enzymes in lipid metabolism pathways to diabetes risk
- Compare lipid profiles between healthy and diseased samples
Example prompts for the agent
- Identify lipid Cer(d18:1/16:0) and retrieve its sphingolipid biosynthesis pathway
- What genes and biomarkers are associated with altered sphingolipid metabolism in neurodegeneration?
- Search HMDB and KEGG for lipid biomarkers of cardiovascular disease
How to install tooluniverse-lipidomics
1Skills CLI (Vercel)
npx skills add mims-harvard/ToolUniverse --skill tooluniverse-lipidomics -a claude-code2Manually: put the files in place
~/.claude/skills/tooluniverse-lipidomics/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/tooluniverse-lipidomics/
1Skills CLI (Vercel)
npx skills add mims-harvard/ToolUniverse --skill tooluniverse-lipidomics -a codex2Manually: put the files in place
.agents/skills/tooluniverse-lipidomics/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add mims-harvard/ToolUniverse --skill tooluniverse-lipidomics -a cursor2Manually: put the files in place
.cursor/skills/tooluniverse-lipidomics/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install mims-harvard/ToolUniverse skills/tooluniverse-lipidomics2Manually: put the files in place
.github/skills/tooluniverse-lipidomics/SKILL.md1Skills CLI (Vercel)
npx skills add mims-harvard/ToolUniverse --skill tooluniverse-lipidomics -a gemini-cli2Platform CLI
gemini skills install https://github.com/mims-harvard/ToolUniverse --path skills/tooluniverse-lipidomicsManually: put the files in place
.devin/skills/tooluniverse-lipidomics/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add mims-harvard/ToolUniverse --skill tooluniverse-lipidomics -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-lipidomics
How do I install tooluniverse-lipidomics in Claude Code?
Run npx skills add mims-harvard/ToolUniverse --skill tooluniverse-lipidomics -a claude-code in your terminal, then verify by opening ~/.claude/skills/tooluniverse-lipidomics/SKILL.md.
What does tooluniverse-lipidomics do?
The skill identifies and classifies lipids using LIPID MAPS categories, retrieves their metabolic pathways (synthesis and degradation enzymes), and finds disease associations via scientific databases. Use it to analyze lipidomics data, discover lipid biomarkers, and explore lipid metabolism in disease.
When should I use tooluniverse-lipidomics instead of other skills?
Use this skill for lipid-specific analysis and disease links. For broad metabolomics (amino acids, sugars, organic acids), use tooluniverse-metabolomics. For drug ADMET prediction, use tooluniverse-admet-prediction.
Source text
SKILL.md · Apache-2.0
---
name: tooluniverse-lipidomics
description: Lipid analysis and lipid-disease associations using LIPID MAPS classification, HMDB metabolite data, KEGG/Reactome lipid pathways (sphingolipid, eicosanoid, steroid, fatty acid), and PubChem chemical info. Use for lipid identification, lipid metabolism pathway mapping, and lipid-associated disease analysis (cardiovascular, diabetes, NAFLD).
disable-model-invocation: true
---
# Lipidomics Analysis
Integrated pipeline for lipid identification, classification, pathway mapping, and disease association analysis. Distinct from general metabolomics because lipids have unique classification systems (LIPID MAPS), specialized pathways (sphingolipid, eicosanoid, steroid), and disease associations (cardiovascular, neurodegeneration, metabolic syndrome).
## Reasoning Strategy
Lipid identification starts with mass spectrometry: the lipid class is determined by the head group fragment mass (e.g., m/z 184 for phosphocholine in positive mode), total chain length and saturation from the precursor exact mass, and individual fatty acid chains from neutral loss or product ion scans. LIPID MAPS classification organizes lipids by chemical structure into 8 categories — knowing the category immediately tells you the likely biological context (sphingolipids → apoptosis/neurodegeneration; glycerophospholipids → membrane remodeling; eicosanoids → inflammation). Structural specificity matters biologically: Cer(d18:1/16:0) and Cer(d18:1/24:1) have different membrane properties and disease associations despite being the same lipid class. Always map changed lipids back to metabolic pathways because lipids are intermediates — an elevated ceramide could mean increased synthesis (CERS activity up), decreased degradation (ASAH1 down), or shunting from sphingomyelin (SMPD1 up).
**LOOK UP DON'T GUESS**: Do not assume a lipid's LIPID MAPS ID, exact mass, or pathway membership — query `LipidMaps_search_by_name` first. Do not guess which diseases are associated with a lipid class; retrieve them from HMDB or CTD.
**Key principles**:
1. **LIPID MAPS classification first** — use the 8-category system (fatty acyls, glycerolipids, glycerophospholipids, sphingolipids, sterol lipids, prenol lipids, saccharolipids, polyketides)
2. **Structural specificity matters** — chain length, unsaturation, and sn-position affect biological function
3. **Connect to pathways** — lipids are metabolic intermediates; always map to biosynthesis/degradation pathways
4. **Disease context** — many lipids are disease biomarkers (sphingolipids in neurodegeneration, oxidized lipids in CVD)
5. **Evidence grading** — T1: clinical biomarker studies, T2: mechanistic studies, T3: association data, T4: computational prediction
---
## When to Use
- "Identify this lipid species from m/z and retention time"
- "What pathways involve ceramide/sphingomyelin?"
- "Lipid biomarkers for Alzheimer's disease"
- "What diseases are associated with altered sphingolipid metabolism?"
- "Map my lipidomics results to KEGG pathways"
- "Compare lipid profiles between conditions"
**Not this skill**: For general metabolomics (amino acids, sugars, organic acids), use `tooluniverse-metabolomics`. For drug ADMET properties, use `tooluniverse-admet-prediction`.
---
## Core Tools
| Tool | Use For |
|------|---------|
| `LipidMaps_search_by_name` | Lipid identification by name, abbreviation, or mass |
| `LipidMaps_get_compound_by_id` | Detailed lipid info (structure, classification, pathways) |
| `SwissLipids_search` | Lipid identification by name/shorthand — SIB's complementary lipid database, use when LIPID MAPS misses a species (see below) |
| `SwissLipids_get_lipid` | Full SwissLipids entry: name, formula, monoisotopic mass, cross-refs |
| `SwissLipids_get_children` | Descend the SwissLipids classification tree (category → class → species) one level |
| `HMDB_search` / `HMDB_get_metabolite` | Lipid metabolite details, disease associations |
| `kegg_search_pathway` | Lipid metabolism pathways (keyword=`sphingolipid`, `glycerolipid`, etc.) |
| `KEGG_get_pathway_genes` | Enzymes in lipid pathways |
| `PubChem_get_compound_properties_by_CID` | Chemical properties (mass, formula, SMILES) |
| `CTD_get_gene_diseases` | Gene-disease links for lipid metabolism enzymes |
| `LipidMaps_get_gene` | Resolve a lipid-metabolism gene (symbol/NCBI ID/LMP ID) to its LIPID MAPS Proteome Database record — name, synonyms, chromosome location, species |
| `LipidMaps_get_protein` | Same lookup at the protein level — adds UniProt/RefSeq cross-references and amino-acid sequence |
| `DisGeNET_search_gene` | Disease associations for lipid genes |
| `PubMed_search_articles` | Published lipidomics studies |
| `OpenTargets_get_associated_drugs_by_target_ensemblID` | Drugs targeting lipid metabolism enzymes |
---
## Workflow
```
Phase 0: Lipid Identity Resolution
Name/mass/abbreviation → LIPID MAPS ID → classification
|
Phase 1: Structural Classification
LIPID MAPS 8-category system → subclass → molecular species
|
Phase 2: Pathway Mapping
KEGG lipid metabolism → biosynthesis/degradation enzymes
|
Phase 3: Disease Associations
CTD/DisGeNET/HMDB → lipid-disease links with evidence
|
Phase 4: Interpretation & Report
Biological significance → biomarker potential → recommendations
```
### Phase 0: Lipid Identity Resolution
```
LipidMaps_search_by_name(query="ceramide") → LMSP ID, exact mass, classification
HMDB_search(compound_name="ceramide") → HMDB ID, disease links
PubChem_get_CID_by_compound_name(name="ceramide") → CID, SMILES
```
**LIPID MAPS search tips**:
- Generic names work well: "ceramide", "sphingomyelin", "phosphatidylcholine"
- Species-level abbreviations like "Cer(d18:1/16:0)" may return 0 results — use the generic class name first, then filter by chain length from results
- For exact mass search: use `LipidMaps_search_by_formula` with molecular formula (e.g., "C34H67NO3")
- If name search fails, try PubChem: `PubChem_get_CID_by_compound_name(name="C16 Ceramide")` then cross-reference
**SwissLipids as a second lipid database** — when LIPID MAPS doesn't have a species
or you want its independent classification, SwissLipids (swisslipids.org, SIB)
covers the same ground with its own ID space and hierarchy:
```
SwissLipids_search(query="PC(16:0/18:1)") → real hit: entity_id "SLM:000088148"
SwissLipids_get_lipid(entity_id="SLM:000000510") → real entry: "hexadecanoate"
(palmitate), formula C16H31O2, mass 255.4167, charge -1, plus a full
adduct_mz table ([M+H]+, [M+Na]+, [M-H]-, etc.) useful for matching
observed MS peaks directly — LIPID MAPS entries don't include this table.
SwissLipids_get_children(entity_id="SLM:000000338") → descend one level in the
classification tree (e.g. from a lipid category down to its member species)
```
Shorthand abbreviations like `PC(16:0/18:1)` work directly in `SwissLipids_search`
(unlike LIPID MAPS, which often needs the generic class name first). Note
`classification` can come back as an empty array `[]` for some entries — a real
data gap in SwissLipids, not a failed lookup; fall back to LIPID MAPS for that
lipid's category if classification is required.
### Phase 1: Structural Classification
Use `LipidMaps_get_compound_by_id` to retrieve the LIPID MAPS 8-category classification (FA, GL, GP, SP, ST, PR, SL, PK) for any lipid. The category immediately signals biological context: SP (sphingolipids) → apoptosis/neurodegeneration; GP (glycerophospholipids) → membrane remodeling; FA-derived eicosanoids → inflammation.
### Phase 2: Pathway Mapping
Key lipid metabolism pathways in KEGG:
| Pathway | KEGG ID | Key Enzymes | Disease Relevance |
|---------|---------|-------------|-------------------|
| Sphingolipid metabolism | hsa00600 | SMPD1, CERS1-6, ASAH1 | Niemann-Pick, Fabry, Gaucher |
| Glycerophospholipid metabolism | hsa00564 | PLA2, LPCAT, LPIN | Barth syndrome, atherosclerosis |
| Arachidonic acid metabolism | hsa00590 | COX1/2, LOX, CYP450 | Inflammation, asthma, CVD |
| Steroid biosynthesis | hsa00100 | HMGCR, CYP51A1, DHCR7 | Hypercholesterolemia, Smith-Lemli-Opitz |
| Fatty acid biosynthesis | hsa00061 | FASN, ACC, SCD | Obesity, NAFLD, cancer |
| Fatty acid degradation | hsa00071 | CPT1, ACADM, HADHA | MCAD deficiency, VLCAD deficiency |
| Bile acid biosynthesis | hsa00120 | CYP7A1, CYP27A1 | Cholestasis, gallstones |
| Ether lipid metabolism | hsa00565 | AGPS, GNPAT | Rhizomelic chondrodysplasia |
```python
# Map lipids to pathways
kegg_search_pathway(keyword="sphingolipid") # → hsa00600
KEGG_get_pathway_genes(pathway_id="hsa00600") # → SMPD1, CERS1, ...
```
### Phase 3: Disease Associations
For each lipid or lipid enzyme, resolve the gene/protein record first if needed, then check disease links:
```python
LipidMaps_get_gene(input_value="SMPD1", input_item="gene_symbol") # LMPD record: name, synonyms, chromosome
# verified live example: input_value="FASN" → {"gene_symbol": "Fasn", "lmp_id": "LMP001577", "gene_name": "fatty acid synthase", ...}
LipidMaps_get_protein(input_value="P49327", input_item="uniprot_id") # adds UniProt/RefSeq xrefs + sequence
CTD_get_gene_diseases(input_terms="SMPD1") # sphingomyelinase → Niemann-Pick
DisGeNET_search_gene(gene="SMPD1") # broader disease associations
HMDB_get_metabolite(compound_name="ceramide") # metabolite-disease links
PubMed_search_articles(query="ceramide biomarker Alzheimer") # clinical evidence
```
**Disease context**: Ceramide elevation → apoptosis, Alzheimer's, insulin resistance. Sphingomyelin depletion → Niemann-Pick. Oxidized phospholipids → CVD. Altered bile acid ratios → NAFLD, cholestasis. Eicosanoid elevation → inflammation. Always verify via HMDB or CTD rather than relying on memory.
### Phase 4: Interpretation & Report
**Computational procedure: Lipid class enrichment analysis**
```python
# When user provides a list of significantly changed lipids
import pandas as pd
from scipy.stats import fisher_exact
# Input: list of changed lipids with LIPID MAPS categories
changed = pd.DataFrame({
'lipid': ['Cer(d18:1/16:0)', 'SM(d18:1/16:0)', 'PC(16:0/18:1)', 'LPC(18:0)'],
'category': ['SP', 'SP', 'GP', 'GP'],
'direction': ['up', 'down', 'unchanged', 'up'],
'fold_change': [2.1, 0.5, 1.1, 1.8]
})
# Count changed vs unchanged per category
from collections import Counter
changed_cats = Counter(changed[changed['direction'] != 'unchanged']['category'])
total_cats = Counter(changed['category'])
# Report
print("Lipid class enrichment:")
for cat in total_cats:
n_changed = changed_cats.get(cat, 0)
n_total = total_cats[cat]
print(f" {cat}: {n_changed}/{n_total} changed")
# Interpretation
if changed_cats.get('SP', 0) / max(total_cats.get('SP', 1), 1) > 0.5:
print("→ Sphingolipid metabolism is significantly altered")
print(" Consider: apoptosis, neurodegeneration, insulin resistance")
```
**Report structure**:
1. **Lipid Identity** — LIPID MAPS classification, structural features
2. **Pathway Context** — which metabolic pathways are affected
3. **Disease Associations** — evidence-graded disease links
4. **Biomarker Assessment** — clinical utility of identified lipid changes
5. **Mechanistic Model** — how lipid changes connect to disease biology
6. **Recommendations** — validation experiments, clinical follow-up
---
## Limitations
- **No raw MS data processing** — this skill interprets identified lipids, not raw spectra. Use LipidSearch, MS-DIAL, or LipiDex for identification first.
- **LIPID MAPS coverage** — some rare or novel lipid species may not be in the database
- **Quantitative thresholds** — fold-change cutoffs are context-dependent; the skill provides frameworks, not universal thresholds
- **Species-specific** — most disease data is human; rat/mouse lipid metabolism can differ significantly
© mims-harvard · Apache-2.0 · mims-harvard/ToolUniverse