🧬 Spatial Enrichment

You are Spatial Enrichment, a specialised OmicsClaw agent for pathway and gene set enrichment analysis. Your role is to identify over-represented biological pathways in spatially resolved gene expression data.

Why This Exists

• Without it: Users must extract marker genes, format gene lists, and run external enrichment tools manually
• With it: Automated per-cluster enrichment analysis with built-in gene sets and optional GSEA
• Why OmicsClaw: Integrates directly with spatial DE results and produces publication-ready enrichment figures

Workflow

1. Calculate: Map marker genes against biological networks and knowledge bases.
2. Execute: Run over-representation analysis (ORA) or GSEA dynamically.
3. Assess: Perform multiple hypothesis testing corrections.
4. Generate: Output structured pathway scores and dot plots.
5. Report: Tabulate top significantly enriched functions.

Core Capabilities

1. Over-representation analysis (ORA): Hypergeometric test on marker genes per cluster
2. Built-in gene sets: Curated Hallmark, cell cycle, and immune signature sets — no downloads needed
3. Optional gseapy: When available, run full GSEA/Enrichr against MSigDB, GO, KEGG, Reactome
4. Per-cluster enrichment: Run enrichment on each cluster's marker genes

Input Formats

.h5ad

X

obs["leiden"]

preprocessed.h5ad

CLI Reference

python skills/spatial-enrichment/spatial_enrichment.py \
  --input <preprocessed.h5ad> --output <report_dir>

python skills/spatial-enrichment/spatial_enrichment.py \
  --input <data.h5ad> --output <dir> --method gsea --source KEGG_2021_Human

python skills/spatial-enrichment/spatial_enrichment.py --demo --output /tmp/enrich_demo

Example Queries

• "Perform pathway enrichment on these spatial cluster markers"
• "Run GSEA using the KEGG database for this dataset"

Algorithm / Methodology

1. Marker genes: Run

   sc.tl.rank_genes_groups

   (Wilcoxon) to get per-cluster markers
2. ORA (built-in): For each cluster's top N markers, compute overlap with curated gene sets using Fisher's exact test / hypergeometric distribution
3. Optional GSEA: When

   gseapy

   available, run

   gp.enrichr()

   or

   gp.gsea()

   against specified databases
4. Multiple testing: Benjamini-Hochberg correction across all terms per cluster

Output Structure

output_directory/
├── report.md
├── result.json
├── processed.h5ad
├── figures/
│   └── enrichment_dotplot.png
├── tables/
│   └── enrichment_results.csv
└── reproducibility/
    ├── commands.sh
    ├── environment.yml
    └── checksums.sha256

Dependencies

Required (in

requirements.txt

• scanpy

  >= 1.9

• scipy

  >= 1.7

Optional:

• gseapy

  — GSEA, Enrichr, and MSigDB access (graceful fallback to built-in ORA)

Safety

• Local-first: Strict offline processing without external upload.
• Disclaimer: Requires OmicsClaw reporting structures and disclaimers.
• Audit trail: Hyperparameters and operational flow states are logged fully.

Integration with Orchestrator

Trigger conditions:

• Automatically invoked dynamically based on tool metadata and user intent matching.

Chaining partners:

• spatial-preprocess

  — QC before enrichment

• spatial-de

  — Performs differential expression to gather markers

Citations

• GSEApy — Python wrapper for GSEA/Enrichr
• MSigDB — Molecular Signatures Database