Spatial Multi-Omics Analysis Pipeline
Comprehensive biological interpretation of spatial omics data. Transforms spatially variable genes (SVGs), domain annotations, and tissue context into actionable biological insights.
KEY PRINCIPLES:
- Report-first approach - Create report file FIRST, then populate progressively
- Domain-by-domain analysis - Characterize each spatial region independently before comparison
- Gene-list-centric - Analyze user-provided SVGs and marker genes with ToolUniverse databases
- Biological interpretation - Go beyond statistics to explain biological meaning of spatial patterns
- Disease focus - Emphasize disease mechanisms and therapeutic opportunities when disease context is provided
- Evidence grading - Grade all evidence as T1 (human/clinical) to T4 (computational)
- Multi-modal thinking - Integrate RNA, protein, and metabolite information when available
- Validation guidance - Suggest experimental validation approaches for key findings
- Source references - Every statement must cite tool/database source
- English-first queries - Always use English terms in tool calls
LOOK UP, DON'T GUESS
When uncertain about any scientific fact, SEARCH databases first (PubMed, UniProt, ChEMBL, ClinVar, etc.) rather than reasoning from memory. A database-verified answer is always more reliable than a guess.
COMPUTE, DON'T DESCRIBE
When analysis requires computation (statistics, data processing, scoring, enrichment), write and run Python code via Bash. Don't describe what you would do — execute it and report actual results. Use ToolUniverse tools to retrieve data, then Python (pandas, scipy, statsmodels, matplotlib) to analyze it.
When to Use This Skill
Apply when users:
- Provide spatially variable genes from spatial transcriptomics experiments
- Ask about biological interpretation of spatial domains/clusters
- Need pathway enrichment of spatial gene expression data
- Want to understand cell-cell interactions from spatial data
- Ask about tumor microenvironment heterogeneity from spatial omics
- Need druggable targets in specific spatial regions
- Ask about tissue zonation patterns (liver, brain, kidney)
- Want to integrate spatial transcriptomics + proteomics data
NOT for: Single gene interpretation (use target-research), variant interpretation, drug safety, bulk RNA-seq, GWAS analysis.
Input Parameters
| Parameter |
Required |
Description |
Example |
| svgs |
Yes |
Spatially variable genes |
['EGFR', 'CDH1', 'VIM', 'MYC', 'CD3E'] |
| tissue_type |
Yes |
Tissue/organ type |
brain, liver, lung, breast |
| technology |
No |
Spatial omics platform |
10x Visium, MERFISH, DBiTplus |
| disease_context |
No |
Disease if applicable |
breast cancer, Alzheimer disease |
| spatial_domains |
No |
Domain -> marker genes dict |
{'Tumor core': ['MYC','EGFR']} |
| cell_types |
No |
Cell types from deconvolution |
['Epithelial', 'T cell'] |
| proteins |
No |
Proteins detected (multi-modal) |
['CD3', 'PD-L1', 'Ki67'] |
| metabolites |
No |
Metabolites (SpatialMETA) |
['glutamine', 'lactate'] |
Spatial Omics Integration Score (0-100)
Data Completeness (0-30): SVGs (5), Disease context (5), Spatial domains (5), Cell types (5), Multi-modal (5), Literature (5)
Biological Insight (0-40): Pathway enrichment FDR<0.05 (10), Cell-cell interactions (10), Disease mechanism (10), Druggable targets (10)
Evidence Quality (0-30): Cross-database validation 3+ DBs (10), Clinical validation (10), Literature support (10)
| Score |
Tier |
Interpretation |
| 80-100 |
Excellent |
Comprehensive characterization, strong insights, druggable targets |
| 60-79 |
Good |
Good pathway/interaction analysis, some therapeutic context |
| 40-59 |
Moderate |
Basic enrichment, limited domain comparison |
| 0-39 |
Limited |
Minimal data, gene-level annotation only |
Evidence Grading
| Tier |
Criteria |
Examples |
| [T1] |
Direct human/clinical evidence |
FDA-approved drug, validated biomarker |
| [T2] |
Experimental evidence |
Validated spatial pattern, known L-R pair |
| [T3] |
Computational/database evidence |
PPI prediction, pathway enrichment |
| [T4] |
Annotation/prediction only |
GO annotation, text-mined association |
Analysis Phases Overview
Phase 0: Input Processing & Disambiguation (ALWAYS FIRST)
Resolve tissue/disease identifiers, establish analysis context. Get MONDO/EFO IDs for disease queries.
- Tools:
OpenTargets_get_disease_id_description_by_name, OpenTargets_get_disease_description_by_efoId, HPA_search_genes_by_query
Phase 1: Gene Characterization
Resolve gene IDs, annotate functions, tissue specificity, subcellular localization.
- Tools:
MyGene_query_genes, UniProt_get_function_by_accession, HPA_get_subcellular_location, HPA_get_rna_expression_by_source, HPA_get_comprehensive_gene_details_by_ensembl_id, HPA_get_cancer_prognostics_by_gene, UniProtIDMap_gene_to_uniprot
Phase 2: Pathway & Functional Enrichment
Identify enriched pathways globally and per-domain. Filter FDR < 0.05.
- Tools:
STRING_functional_enrichment (PRIMARY), ReactomeAnalysis_pathway_enrichment, GO_get_annotations_for_gene, kegg_search_pathway, WikiPathways_search
Phase 3: Spatial Domain Characterization
Characterize each domain biologically, assign cell types from markers, compare domains.
- Tools: Phase 2 tools +
HPA_get_biological_processes_by_gene, HPA_get_protein_interactions_by_gene
Phase 4: Cell-Cell Interaction Inference
Predict communication from spatial patterns. Check ligand-receptor pairs across domains.
- Tools:
STRING_get_interaction_partners, STRING_get_protein_interactions, intact_search_interactions, Reactome_get_interactor, DGIdb_get_drug_gene_interactions
Phase 5: Disease & Therapeutic Context
Connect to disease mechanisms, identify druggable targets, find clinical trials.
- Tools:
OpenTargets_get_associated_targets_by_disease_efoId, OpenTargets_get_target_tractability_by_ensemblID, OpenTargets_get_associated_drugs_by_target_ensemblID, search_clinical_trials, DGIdb_get_gene_druggability, civic_search_genes
Phase 6: Multi-Modal Integration
Integrate protein/RNA/metabolite data. Compare spatial RNA with protein detection.
- Tools:
HPA_get_subcellular_location, HPA_get_rna_expression_in_specific_tissues, Reactome_map_uniprot_to_pathways, kegg_get_pathway_info
Phase 7: Immune Microenvironment (Cancer/Inflammation only)
Classify immune cells, check checkpoint expression, assess Hot vs Cold vs Excluded patterns.
- Tools:
STRING_functional_enrichment, OpenTargets_get_target_tractability_by_ensemblID, iedb_search_epitopes
Phase 8: Literature & Validation Context
Search published evidence, suggest validation experiments (smFISH, IHC, PLA).
- Tools:
PubMed_search_articles, openalex_literature_search
Data Discovery: HuBMAP Spatial Atlas Tools
Use HuBMAP tools to find published spatial biology reference datasets for comparison, validation, or cross-study analysis.
| Tool |
Purpose |
Key Parameters |
HuBMAP_search_datasets |
Search published spatial datasets by organ/assay/keyword |
organ (code: "LK"=Kidney, "BR"=Brain, "LU"=Lung, etc.), dataset_type ("RNAseq", "CODEX", "MALDI"), query, limit |
HuBMAP_list_organs |
List all available organs with codes and UBERON IDs |
(no required params) |
HuBMAP_get_dataset |
Get detailed metadata for a specific HuBMAP dataset |
hubmap_id (e.g. "HBM626.FHJD.938") |
When to use: Phase 0 (find reference datasets for the tissue), Phase 8 (cross-reference findings with published HuBMAP atlas data).
See phase-procedures.md for detailed workflows, decision logic, and tool parameter specifications per phase.
Report Structure
Create file: {tissue}_{disease}_spatial_omics_report.md
# Spatial Multi-Omics Analysis Report: {Tissue Type}
**Report Generated**: {date} | **Technology**: {platform}
**Tissue**: {tissue_type} | **Disease**: {disease or "Normal tissue"}
**Total SVGs**: {count} | **Spatial Domains**: {count}
**Spatial Omics Integration Score**: (calculated after analysis)
## Executive Summary
## 1. Tissue & Disease Context
## 2. Spatially Variable Gene Characterization
- 2.1 Gene ID Resolution
- 2.2 Tissue Expression Patterns
- 2.3 Subcellular Localization
- 2.4 Disease Associations
## 3. Pathway Enrichment Analysis
- 3.1 STRING, 3.2 Reactome, 3.3-3.5 GO (BP, MF, CC)
## 4. Spatial Domain Characterization (per-domain + comparison)
## 5. Cell-Cell Interaction Inference
- 5.1 PPI, 5.2 Ligand-Receptor, 5.3 Signaling Pathways
## 6. Disease & Therapeutic Context
- 6.1 Disease Gene Overlap, 6.2 Druggable Targets, 6.3 Drug Mechanisms, 6.4 Trials
## 7. Multi-Modal Integration (if data available)
## 8. Immune Microenvironment (if relevant)
## 9. Literature & Validation Context
## Spatial Omics Integration Score (breakdown table)
## Completeness Checklist
## References (tools used, database versions)
See report-template.md for full template with table structures.
Completeness Checklist
Common Use Cases
- Cancer Spatial Heterogeneity: Visium with tumor/stroma/immune domains -> pathways, immune infiltration, druggable targets, checkpoints
- Brain Tissue Zonation: MERFISH with neuronal subtypes -> synaptic signaling, receptors, hippocampal zonation
- Liver Metabolic Zonation: Periportal vs pericentral -> CYP450, Wnt gradient, drug metabolism enzymes
- Tumor-Immune Interface: DBiTplus RNA+protein -> checkpoint L-R pairs, immune exclusion, multi-modal concordance
- Developmental Patterns: Morphogen gradients (Wnt, BMP, FGF, SHH), TF patterns, cell fate genes
- Disease Progression: Disease gradient -> inflammatory response, neuronal loss, therapeutic windows
Reference Files
- phase-procedures.md - Detailed phase workflows, decision logic, tool usage per phase
- tool-reference.md - Tool parameter names, response formats, fallback strategies, limitations
- reference-data.md - Cell type markers, ligand-receptor pairs, immune checkpoint reference
- report-template.md - Full report template with all table structures
- test_spatial_omics.py - Test suite
Summary
Spatial Multi-Omics Analysis provides:
- Gene characterization (ID resolution, function, localization, tissue expression)
- Pathway & functional enrichment (STRING, Reactome, GO, KEGG)
- Spatial domain characterization (per-domain and cross-domain)
- Cell-cell interaction inference (PPI, ligand-receptor, signaling)
- Disease & therapeutic context (disease genes, druggable targets, trials)
- Multi-modal integration (RNA-protein concordance, metabolic pathways)
- Immune microenvironment (cell types, checkpoints, immunotherapy)
- Literature context & validation recommendations
Outputs: Markdown report with Spatial Omics Integration Score (0-100)
Uses: 70+ ToolUniverse tools across 9 analysis phases
Time: ~10-20 minutes depending on gene list size
1---23name: tooluniverse-spatial-omics-analysis-24description: "Spatial multi-omics interpretation pipeline. Transforms spatially variable genes (SVGs), domain annotations, and tissue context into biological insights via domain-by-domain characterization, cell-type composition, spatial gene expression patterns, RNA+protein+metabolite integration. Use for Visium, MERFISH, seqFISH, Slide-seq, spatial proteomics, and spatial multi-omics interpretation. Goes beyond statistics to disease mechanisms and therapeutic opportunities."5metadata:6 source: https://skillmd.com/skills/mims-harvard/tooluniverse-spatial-omics-analysis-27---89# Spatial Multi-Omics Analysis Pipeline1011Comprehensive biological interpretation of spatial omics data. Transforms spatially variable genes (SVGs), domain annotations, and tissue context into actionable biological insights.1213**KEY PRINCIPLES**:141. **Report-first approach** - Create report file FIRST, then populate progressively152. **Domain-by-domain analysis** - Characterize each spatial region independently before comparison163. **Gene-list-centric** - Analyze user-provided SVGs and marker genes with ToolUniverse databases174. **Biological interpretation** - Go beyond statistics to explain biological meaning of spatial patterns185. **Disease focus** - Emphasize disease mechanisms and therapeutic opportunities when disease context is provided196. **Evidence grading** - Grade all evidence as T1 (human/clinical) to T4 (computational)207. **Multi-modal thinking** - Integrate RNA, protein, and metabolite information when available218. **Validation guidance** - Suggest experimental validation approaches for key findings229. **Source references** - Every statement must cite tool/database source2310. **English-first queries** - Always use English terms in tool calls2425---2627## LOOK UP, DON'T GUESS28When uncertain about any scientific fact, SEARCH databases first (PubMed, UniProt, ChEMBL, ClinVar, etc.) rather than reasoning from memory. A database-verified answer is always more reliable than a guess.2930---3132## COMPUTE, DON'T DESCRIBE33When analysis requires computation (statistics, data processing, scoring, enrichment), write and run Python code via Bash. Don't describe what you would do — execute it and report actual results. Use ToolUniverse tools to retrieve data, then Python (pandas, scipy, statsmodels, matplotlib) to analyze it.3435## When to Use This Skill3637Apply when users:38- Provide spatially variable genes from spatial transcriptomics experiments39- Ask about biological interpretation of spatial domains/clusters40- Need pathway enrichment of spatial gene expression data41- Want to understand cell-cell interactions from spatial data42- Ask about tumor microenvironment heterogeneity from spatial omics43- Need druggable targets in specific spatial regions44- Ask about tissue zonation patterns (liver, brain, kidney)45- Want to integrate spatial transcriptomics + proteomics data4647**NOT for**: Single gene interpretation (use target-research), variant interpretation, drug safety, bulk RNA-seq, GWAS analysis.4849---5051## Input Parameters5253| Parameter | Required | Description | Example |54|-----------|----------|-------------|---------|55| **svgs** | Yes | Spatially variable genes | `['EGFR', 'CDH1', 'VIM', 'MYC', 'CD3E']` |56| **tissue_type** | Yes | Tissue/organ type | `brain`, `liver`, `lung`, `breast` |57| **technology** | No | Spatial omics platform | `10x Visium`, `MERFISH`, `DBiTplus` |58| **disease_context** | No | Disease if applicable | `breast cancer`, `Alzheimer disease` |59| **spatial_domains** | No | Domain -> marker genes dict | `{'Tumor core': ['MYC','EGFR']}` |60| **cell_types** | No | Cell types from deconvolution | `['Epithelial', 'T cell']` |61| **proteins** | No | Proteins detected (multi-modal) | `['CD3', 'PD-L1', 'Ki67']` |62| **metabolites** | No | Metabolites (SpatialMETA) | `['glutamine', 'lactate']` |6364---6566## Spatial Omics Integration Score (0-100)6768**Data Completeness (0-30)**: SVGs (5), Disease context (5), Spatial domains (5), Cell types (5), Multi-modal (5), Literature (5)6970**Biological Insight (0-40)**: Pathway enrichment FDR<0.05 (10), Cell-cell interactions (10), Disease mechanism (10), Druggable targets (10)7172**Evidence Quality (0-30)**: Cross-database validation 3+ DBs (10), Clinical validation (10), Literature support (10)7374| Score | Tier | Interpretation |75|-------|------|----------------|76| 80-100 | Excellent | Comprehensive characterization, strong insights, druggable targets |77| 60-79 | Good | Good pathway/interaction analysis, some therapeutic context |78| 40-59 | Moderate | Basic enrichment, limited domain comparison |79| 0-39 | Limited | Minimal data, gene-level annotation only |8081### Evidence Grading8283| Tier | Criteria | Examples |84|------|----------|----------|85| [T1] | Direct human/clinical evidence | FDA-approved drug, validated biomarker |86| [T2] | Experimental evidence | Validated spatial pattern, known L-R pair |87| [T3] | Computational/database evidence | PPI prediction, pathway enrichment |88| [T4] | Annotation/prediction only | GO annotation, text-mined association |8990---9192## Analysis Phases Overview9394### Phase 0: Input Processing & Disambiguation (ALWAYS FIRST)95Resolve tissue/disease identifiers, establish analysis context. Get MONDO/EFO IDs for disease queries.96- Tools: `OpenTargets_get_disease_id_description_by_name`, `OpenTargets_get_disease_description_by_efoId`, `HPA_search_genes_by_query`9798### Phase 1: Gene Characterization99Resolve gene IDs, annotate functions, tissue specificity, subcellular localization.100- Tools: `MyGene_query_genes`, `UniProt_get_function_by_accession`, `HPA_get_subcellular_location`, `HPA_get_rna_expression_by_source`, `HPA_get_comprehensive_gene_details_by_ensembl_id`, `HPA_get_cancer_prognostics_by_gene`, `UniProtIDMap_gene_to_uniprot`101102### Phase 2: Pathway & Functional Enrichment103Identify enriched pathways globally and per-domain. Filter FDR < 0.05.104- Tools: `STRING_functional_enrichment` (PRIMARY), `ReactomeAnalysis_pathway_enrichment`, `GO_get_annotations_for_gene`, `kegg_search_pathway`, `WikiPathways_search`105106### Phase 3: Spatial Domain Characterization107Characterize each domain biologically, assign cell types from markers, compare domains.108- Tools: Phase 2 tools + `HPA_get_biological_processes_by_gene`, `HPA_get_protein_interactions_by_gene`109110### Phase 4: Cell-Cell Interaction Inference111Predict communication from spatial patterns. Check ligand-receptor pairs across domains.112- Tools: `STRING_get_interaction_partners`, `STRING_get_protein_interactions`, `intact_search_interactions`, `Reactome_get_interactor`, `DGIdb_get_drug_gene_interactions`113114### Phase 5: Disease & Therapeutic Context115Connect to disease mechanisms, identify druggable targets, find clinical trials.116- Tools: `OpenTargets_get_associated_targets_by_disease_efoId`, `OpenTargets_get_target_tractability_by_ensemblID`, `OpenTargets_get_associated_drugs_by_target_ensemblID`, `search_clinical_trials`, `DGIdb_get_gene_druggability`, `civic_search_genes`117118### Phase 6: Multi-Modal Integration119Integrate protein/RNA/metabolite data. Compare spatial RNA with protein detection.120- Tools: `HPA_get_subcellular_location`, `HPA_get_rna_expression_in_specific_tissues`, `Reactome_map_uniprot_to_pathways`, `kegg_get_pathway_info`121122### Phase 7: Immune Microenvironment (Cancer/Inflammation only)123Classify immune cells, check checkpoint expression, assess Hot vs Cold vs Excluded patterns.124- Tools: `STRING_functional_enrichment`, `OpenTargets_get_target_tractability_by_ensemblID`, `iedb_search_epitopes`125126### Phase 8: Literature & Validation Context127Search published evidence, suggest validation experiments (smFISH, IHC, PLA).128- Tools: `PubMed_search_articles`, `openalex_literature_search`129130### Data Discovery: HuBMAP Spatial Atlas Tools131132Use HuBMAP tools to find published spatial biology reference datasets for comparison, validation, or cross-study analysis.133134| Tool | Purpose | Key Parameters |135|------|---------|----------------|136| `HuBMAP_search_datasets` | Search published spatial datasets by organ/assay/keyword | `organ` (code: "LK"=Kidney, "BR"=Brain, "LU"=Lung, etc.), `dataset_type` ("RNAseq", "CODEX", "MALDI"), `query`, `limit` |137| `HuBMAP_list_organs` | List all available organs with codes and UBERON IDs | (no required params) |138| `HuBMAP_get_dataset` | Get detailed metadata for a specific HuBMAP dataset | `hubmap_id` (e.g. "HBM626.FHJD.938") |139140**When to use**: Phase 0 (find reference datasets for the tissue), Phase 8 (cross-reference findings with published HuBMAP atlas data).141142See **phase-procedures.md** for detailed workflows, decision logic, and tool parameter specifications per phase.143144---145146## Report Structure147148Create file: `{tissue}_{disease}_spatial_omics_report.md`149150```151# Spatial Multi-Omics Analysis Report: {Tissue Type}152**Report Generated**: {date} | **Technology**: {platform}153**Tissue**: {tissue_type} | **Disease**: {disease or "Normal tissue"}154**Total SVGs**: {count} | **Spatial Domains**: {count}155**Spatial Omics Integration Score**: (calculated after analysis)156157## Executive Summary158## 1. Tissue & Disease Context159## 2. Spatially Variable Gene Characterization160 - 2.1 Gene ID Resolution161 - 2.2 Tissue Expression Patterns162 - 2.3 Subcellular Localization163 - 2.4 Disease Associations164## 3. Pathway Enrichment Analysis165 - 3.1 STRING, 3.2 Reactome, 3.3-3.5 GO (BP, MF, CC)166## 4. Spatial Domain Characterization (per-domain + comparison)167## 5. Cell-Cell Interaction Inference168 - 5.1 PPI, 5.2 Ligand-Receptor, 5.3 Signaling Pathways169## 6. Disease & Therapeutic Context170 - 6.1 Disease Gene Overlap, 6.2 Druggable Targets, 6.3 Drug Mechanisms, 6.4 Trials171## 7. Multi-Modal Integration (if data available)172## 8. Immune Microenvironment (if relevant)173## 9. Literature & Validation Context174## Spatial Omics Integration Score (breakdown table)175## Completeness Checklist176## References (tools used, database versions)177```178179See **report-template.md** for full template with table structures.180181---182183## Completeness Checklist184185- [ ] Gene ID resolution complete186- [ ] Tissue expression patterns analyzed (HPA)187- [ ] Subcellular localization checked (HPA)188- [ ] Pathway enrichment complete (STRING + Reactome)189- [ ] GO enrichment complete (BP + MF + CC)190- [ ] Spatial domains characterized individually191- [ ] Domain comparison performed192- [ ] PPI analyzed (STRING)193- [ ] Ligand-receptor pairs identified194- [ ] Disease associations checked (OpenTargets)195- [ ] Druggable targets identified196- [ ] Multi-modal integration performed (if data available)197- [ ] Immune microenvironment characterized (if relevant)198- [ ] Literature search completed199- [ ] Validation recommendations provided200- [ ] Integration Score calculated201- [ ] Executive summary written202- [ ] All sections have source citations203204---205206## Common Use Cases2072081. **Cancer Spatial Heterogeneity**: Visium with tumor/stroma/immune domains -> pathways, immune infiltration, druggable targets, checkpoints2092. **Brain Tissue Zonation**: MERFISH with neuronal subtypes -> synaptic signaling, receptors, hippocampal zonation2103. **Liver Metabolic Zonation**: Periportal vs pericentral -> CYP450, Wnt gradient, drug metabolism enzymes2114. **Tumor-Immune Interface**: DBiTplus RNA+protein -> checkpoint L-R pairs, immune exclusion, multi-modal concordance2125. **Developmental Patterns**: Morphogen gradients (Wnt, BMP, FGF, SHH), TF patterns, cell fate genes2136. **Disease Progression**: Disease gradient -> inflammatory response, neuronal loss, therapeutic windows214215---216217## Reference Files218219- **phase-procedures.md** - Detailed phase workflows, decision logic, tool usage per phase220- **tool-reference.md** - Tool parameter names, response formats, fallback strategies, limitations221- **reference-data.md** - Cell type markers, ligand-receptor pairs, immune checkpoint reference222- **report-template.md** - Full report template with all table structures223- **test_spatial_omics.py** - Test suite224225---226227## Summary228229**Spatial Multi-Omics Analysis** provides:2301. Gene characterization (ID resolution, function, localization, tissue expression)2312. Pathway & functional enrichment (STRING, Reactome, GO, KEGG)2323. Spatial domain characterization (per-domain and cross-domain)2334. Cell-cell interaction inference (PPI, ligand-receptor, signaling)2345. Disease & therapeutic context (disease genes, druggable targets, trials)2356. Multi-modal integration (RNA-protein concordance, metabolic pathways)2367. Immune microenvironment (cell types, checkpoints, immunotherapy)2378. Literature context & validation recommendations238239**Outputs**: Markdown report with Spatial Omics Integration Score (0-100)240**Uses**: 70+ ToolUniverse tools across 9 analysis phases241**Time**: ~10-20 minutes depending on gene list size