home / skills / gptomics / bioskills / spatial-proteomics

spatial-proteomics skill

/spatial-transcriptomics/spatial-proteomics

This skill analyzes spatial proteomics data from CODEX, IMC, and MIBI to phenotype cells, assess spatial neighborhoods, and integrate with transcriptomics.

npx playbooks add skill gptomics/bioskills --skill spatial-proteomics

Review the files below or copy the command above to add this skill to your agents.

Files (3)

SKILL.md

3.8 KB

---
name: bio-spatial-transcriptomics-spatial-proteomics
description: Analyzes spatial proteomics data from CODEX, IMC, and MIBI platforms including cell segmentation and protein colocalization. Use when working with multiplexed imaging data, analyzing protein spatial patterns, or integrating spatial proteomics with transcriptomics.
tool_type: python
primary_tool: scimap
---

## Version Compatibility

Reference examples tested with: anndata 0.10+, scanpy 1.10+, squidpy 1.3+

Before using code patterns, verify installed versions match. If versions differ:
- Python: `pip show <package>` then `help(module.function)` to check signatures

If code throws ImportError, AttributeError, or TypeError, introspect the installed
package and adapt the example to match the actual API rather than retrying.

# Spatial Proteomics Analysis

**"Analyze my CODEX/IMC spatial proteomics data"** → Process multiplexed imaging data including cell segmentation, protein phenotyping, spatial neighborhood analysis, and protein colocalization scoring.
- Python: `scimap.tl.phenotype_cells()`, `squidpy.gr.nhood_enrichment()`

## Data Loading

**Goal:** Process multiplexed spatial proteomics data (CODEX/IMC/MIBI) through cell phenotyping, spatial neighborhood analysis, and protein colocalization scoring.

**Approach:** Load the cell-by-marker intensity matrix with spatial coordinates into AnnData, normalize and rescale marker intensities, phenotype cells by marker expression gating, then analyze spatial neighborhoods and cell-cell interactions using scimap and squidpy.

```python
import scimap as sm
import anndata as ad

# Load CODEX/IMC data (cell x marker matrix with spatial coordinates)
adata = ad.read_h5ad('spatial_proteomics.h5ad')

# Required: spatial coordinates in adata.obsm['spatial']
# Required: protein intensities in adata.X
```

## Preprocessing

```python
# Log transform intensities
sm.pp.log1p(adata)

# Rescale markers (0-1 per marker)
sm.pp.rescale(adata)

# Combat batch correction if multiple FOVs
sm.pp.combat(adata, batch_key='fov')
```

## Phenotyping Cells

```python
# Manual gating approach
phenotype_markers = {
    'T_cell': ['CD3', 'CD45'],
    'B_cell': ['CD20', 'CD45'],
    'Macrophage': ['CD68', 'CD163'],
    'Tumor': ['panCK', 'Ki67']
}

sm.tl.phenotype_cells(adata, phenotype=phenotype_markers,
                      gate=0.5, label='phenotype')

# Clustering-based phenotyping
sm.tl.cluster(adata, method='leiden', resolution=1.0)
```

## Spatial Analysis

```python
# Build spatial neighbors graph
sm.tl.spatial_distance(adata, x_coordinate='X', y_coordinate='Y')

# Neighborhood enrichment
sm.tl.spatial_interaction(adata, phenotype='phenotype',
                          method='knn', knn=10)

# Spatial clustering (communities of cells)
sm.tl.spatial_cluster(adata, phenotype='phenotype')
```

## Visualization

```python
# Spatial scatter plot
sm.pl.spatial_scatterPlot(adata, colorBy='phenotype',
                          x='X', y='Y', s=5)

# Heatmap of spatial interactions
sm.pl.spatial_interaction(adata)

# Marker expression overlay
sm.pl.image_viewer(adata, markers=['CD3', 'CD20', 'panCK'])
```

## Integration with Transcriptomics

```python
import squidpy as sq

# If matched spatial transcriptomics available
# Transfer labels or integrate modalities
sq.gr.spatial_neighbors(adata_protein)
sq.gr.spatial_neighbors(adata_rna)

# Compare spatial patterns across modalities
```

## Platform-Specific Notes

| Platform | Markers | Resolution | Notes |
|----------|---------|------------|-------|
| CODEX | 40-60 | Subcellular | Cyclic staining |
| IMC | 40+ | 1 um | Metal-tagged antibodies |
| MIBI | 40+ | 260 nm | Mass spectrometry |

## Related Skills

- spatial-transcriptomics/spatial-neighbors - Spatial graph construction
- spatial-transcriptomics/spatial-domains - Domain identification
- imaging-mass-cytometry/phenotyping - IMC-specific analysis

Overview

This skill analyzes spatial proteomics data from CODEX, IMC, and MIBI to produce cell segmentations, phenotype assignments, spatial neighborhood statistics, and protein colocalization scores. It targets multiplexed imaging datasets loaded into AnnData and integrates spatial proteomics with spatial transcriptomics when available. The outputs are actionable cell labels, interaction matrices, and visualizations for spatial biology interpretation.

How this skill works

Load a cell-by-marker intensity matrix and spatial coordinates into AnnData, normalize and rescale marker intensities, and optionally perform batch correction. Phenotype cells via marker gating or clustering, build spatial neighbor graphs, compute neighborhood enrichment and spatial clusters, and score protein colocalization. The workflow uses scimap for proteomics-centric preprocessing and phenotyping, and squidpy for spatial neighbor construction and cross-modality comparisons.

When to use it

Working with multiplexed imaging data from CODEX, IMC, or MIBI.
You need per-cell protein phenotype calls from imaging data.
Studying cell–cell neighborhoods, tissue microenvironments, or spatial interaction enrichment.
Comparing spatial protein patterns to matched spatial transcriptomics.
Generating spatial visualizations of marker expression and interaction heatmaps.

Best practices

Ensure spatial coordinates are present in adata.obsm['spatial'] and intensities in adata.X before processing.
Log-transform and rescale marker intensities per-marker to stabilize comparisons and downstream gating.
If multiple fields of view or batches exist, run batch correction (e.g., Combat) prior to phenotype calling.
Validate automated phenotypes with manual gating and marker overlays; adjust gating thresholds per dataset.
Inspect package versions (anndata, scanpy, squidpy, scimap) and adapt API calls if signatures differ.

Example use cases

Phenotype tumor and immune populations in CODEX data, then quantify immune cell neighborhood enrichment around tumor regions.
Compute protein colocalization scores in IMC to identify spatially co-expressed signaling complexes.
Integrate MIBI-derived protein phenotypes with spatial transcriptomics to transfer labels and compare modality-specific spatial patterns.
Build spatial interaction heatmaps to highlight enriched contacts between macrophages and T cells in inflamed tissue.
Cluster spatial communities to map microanatomic domains (e.g., tumor nests, stromal ribbons).

FAQ

What input format is expected?

A single AnnData object with per-cell marker intensities in adata.X and spatial coordinates in adata.obsm['spatial'].

Which preprocessing steps are essential?

Log1p transform, per-marker rescaling (0–1), and batch correction when multiple FOVs or experiments are present.

Can I combine this with spatial RNA data?

Yes. Use squidpy to compute spatial neighbors for each modality and transfer labels or compare spatial patterns across protein and RNA datasets.