domains / science / spatial-multiomics-integration / spatial-multiomics

spatial-multiomics

Skill card

combined

The two skills use the same name but define different tasks. The first copy is a platform-specific analysis guide for high-resolution spatial transcriptomics (Slide-seq, Stereo-seq, Visium HD, MERFISH) using Squidpy and SpatialData. The second copy is a conceptual and methodological guide for integrating spatial RNA with a second modality (protein, ATAC) using Muon, MOFA, and registration tools like PASTE. They disagree on the primary tool (squidpy vs muon) and the core objective (high-res spatial analysis vs multi-modal integration). Combining them is necessary to capture both the high-resolution processing steps and the multi-modal integration logic.

Repositories

Other copies

Combined skill

---
name: spatial-multiomics
description: Analyze high-resolution spatial platforms and integrate spatial RNA with secondary modalities (protein, ATAC).
tool_type: python
primary_tool: squidpy, muon
---

# Spatial Multi-omics Analysis

## 1. High-Resolution Spatial Analysis (Squidpy)
Use this section for subcellular or high-density data (Slide-seq, Stereo-seq, Visium HD).

```python
import squidpy as sq
import scanpy as sc

# Load spatial data
adata = sc.read_h5ad('spatial_multiomics.h5ad')

# Spatial neighbors
sq.gr.spatial_neighbors(adata, coord_type='generic', n_neighs=10, spatial_key='spatial')

# Spatial autocorrelation
sq.gr.spatial_autocorr(adata, mode='moran', genes=adata.var_names[:100])

# Neighborhood enrichment
sq.gr.nhood_enrichment(adata, cluster_key='cell_type')
```

## 2. Multi-Modal Integration (Muon)
Use this section when integrating spatial RNA with protein or ATAC data.

### Determine Integration Regime
- **Vertical (Same-pixel):** Modalities are co-measured on the same pixels (e.g., spatial-CITE-seq, DBiT-seq). Use joint factor methods like MOFA or WNN.
- **Diagonal (Serial-section):** Modalities are from different sections (e.g., RNA on section N, ATAC on N+1). Use registration methods like PASTE or STalign. Do NOT use same-cell coupling methods.

```python
import muon as mu

# Example: Vertical integration using MOFA
# mdata = mu.MuData({'rna': adata_rna, 'protein': adata_protein})
# mu.tl.mofa(mdata)
```

## 3. Quality Metrics
| Metric | Visium | High-Resolution |
|--------|--------|-----------------|
| Genes/spot | >2000 | >500 |
| UMI/spot | >5000 | >1000 |
| Spatial coverage | >80% | >50% |