Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

Bio Workflows Spatial Pipeline · FreedomIntelligence | DecimalAI

Bio Workflows Spatial Pipeline

SkillScore82Maintained

bio-workflows-spatial-pipeline

End-to-end spatial transcriptomics workflow for Visium/Xenium data. Covers data loading, preprocessing, spatial analysis, domain detection, and visualization with Squidpy. Use when analyzing spatial transcriptomics data.

scienceby FreedomIntelligence

Links this skill to your agents and keeps it current with the author’s updates.

Automatic

Data Analyst pack

1 · One-time — install the CLI (docs ↗)

pip install decimalai

2 · Get the skill — writes .claude/skills/bio-workflows-spatial-pipeline/SKILL.md

decimalai skills pull bio-workflows-spatial-pipeline --out .claude/skills/

No CLI? Get the same file — curl · raw file

curl -fsSL --create-dirs -o .claude/skills/bio-workflows-spatial-pipeline/SKILL.md https://app.decimal.ai/s/bio-workflows-spatial-pipeline/SKILL.md

Raw file: /s/bio-workflows-spatial-pipeline/SKILL.md ↗ · machine-readable summary at /s/bio-workflows-spatial-pipeline.json ↗

In productionGlobal

how this skill behaves on real traffic · last 30 days

No production data for this skill yet — usage numbers appear here once agents start running it.

Anonymized aggregate across every org using this skill — proof from real production use, not a demo.

Does this skill help? · Benchmark: with vs. withoutsign in to run this on your agents

22 test cases from the publisher’s eval.yaml · each run twice on gemini-3.6-flash — with the skill injected, then without · last run 50d ago

Yes — +27% lift · at higher cost+67% tokens0% turns82% of whole cases pass with skill (18/22)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-11	✗→✓	▲ Improved	92%	0%
case-03	✗→✓	▲ Improved	80%	0%
case-06	✗→✓	▲ Improved	55%	0%
case-09	✗→✓	▲ Improved	69%	0%
case-14	✗→✓	▲ Improved	85%	0%

Showing 5 of 22 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/1/2026

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

FreedomIntelligence / OpenClaw-Medical-Skills

skills/bio-workflows-spatial-pipeline/SKILL.md

View on GitHub ↗

Imported 8/1/2026Version v2

Skill contents

SKILL.md

<!--

COPYRIGHT NOTICE

This file is part of the "Universal Biomedical Skills" project.

Copyright (c) 2026 MD BABU MIA, PhD <md.babu.mia@mssm.edu>

All Rights Reserved.

This code is proprietary and confidential.

Unauthorized copying of this file, via any medium is strictly prohibited.

Provenance: Authenticated by MD BABU MIA

-->

Spatial Transcriptomics Pipeline

Complete workflow for analyzing Visium, Xenium, or other spatial transcriptomics data.

Workflow Overview

Spatial data (Space Ranger output)
    |
    v
[1. Load Data] ---------> Read Visium/Xenium
    |
    v
[2. QC & Preprocessing] -> Filter, normalize
    |
    v
[3. Clustering] --------> Standard scRNA-seq clustering
    |
    v
[4. Spatial Analysis] --> Neighbors, statistics
    |
    v
[5. Domain Detection] --> Spatial domains
    |
    v
[6. Visualization] -----> Spatial plots
    |
    v
Annotated spatial data

Primary Path: Squidpy + Scanpy

Step 1: Load Data

python
import scanpy as sc
import squidpy as sq
import numpy as np
import matplotlib.pyplot as plt

# Load Visium data (Space Ranger output)
adata = sq.read.visium('spaceranger_output/')

# Or load from specific files
adata = sc.read_10x_h5('filtered_feature_bc_matrix.h5')
adata.uns['spatial'] = ...  # Add spatial info

# For Xenium
adata = sq.read.xenium('xenium_output/')

print(f'Loaded: {adata.n_obs} spots/cells, {adata.n_vars} genes')

Step 2: Quality Control

python
# QC metrics
adata.var['mt'] = adata.var_names.str.startswith('MT-')
sc.pp.calculate_qc_metrics(adata, qc_vars=['mt'], inplace=True)

# Visualize QC
fig, axes = plt.subplots(1, 3, figsize=(15, 4))
sc.pl.spatial(adata, color='total_counts', ax=axes[0], show=False)
sc.pl.spatial(adata, color='n_genes_by_counts', ax=axes[1], show=False)
sc.pl.spatial(adata, color='pct_counts_mt', ax=axes[2], show=False)
plt.savefig('qc_spatial.pdf')

# Filter
sc.pp.filter_cells(adata, min_counts=500)
sc.pp.filter_cells(adata, min_genes=200)
sc.pp.filter_genes(adata, min_cells=10)
adata = adata[adata.obs.pct_counts_mt < 25, :]

print(f'After QC: {adata.n_obs} spots/cells')

Step 3: Normalization and Clustering

python
# Store raw counts
adata.layers['counts'] = adata.X.copy()

# Normalize
sc.pp.normalize_total(adata, target_sum=1e4)
sc.pp.log1p(adata)

# HVGs
sc.pp.highly_variable_genes(adata, n_top_genes=2000)

# PCA and clustering
adata.raw = adata
adata = adata[:, adata.var.highly_variable]
sc.pp.scale(adata, max_value=10)
sc.tl.pca(adata, n_comps=50)
sc.pp.neighbors(adata, n_neighbors=15, n_pcs=30)
sc.tl.umap(adata)
sc.tl.leiden(adata, resolution=0.5)

# Visualize clusters in space
sc.pl.spatial(adata, color='leiden', spot_size=1.5)
plt.savefig('clusters_spatial.pdf')

Step 4: Spatial Analysis

python
# Build spatial neighbors graph
sq.gr.spatial_neighbors(adata, coord_type='generic', n_neighs=6)

# Neighborhood enrichment (which clusters are neighbors)
sq.gr.nhood_enrichment(adata, cluster_key='leiden')
sq.pl.nhood_enrichment(adata, cluster_key='leiden')
plt.savefig('nhood_enrichment.pdf')

# Co-occurrence analysis
sq.gr.co_occurrence(adata, cluster_key='leiden')
sq.pl.co_occurrence(adata, cluster_key='leiden')
plt.savefig('co_occurrence.pdf')

# Spatially variable genes
sq.gr.spatial_autocorr(adata, mode='moran', n_perms=100, n_jobs=4)

# Top spatially variable genes
svg = adata.uns['moranI'].sort_values('I', ascending=False)
top_svg = svg.head(20).index.tolist()
print('Top spatially variable genes:', top_svg[:10])

Step 5: Domain Detection

python
# Spatial domain detection using clustering with spatial constraints
# Option 1: Use spatial neighbors for Leiden clustering
sq.gr.spatial_neighbors(adata, coord_type='generic', n_neighs=15)
sc.tl.leiden(adata, resolution=0.3, key_added='spatial_domains',
             adjacency=adata.obsp['spatial_connectivities'])

# Visualize domains
sc.pl.spatial(adata, color='spatial_domains', spot_size=1.5)
plt.savefig('spatial_domains.pdf')

# Compare transcriptomic vs spatial clusters
sc.pl.spatial(adata, color=['leiden', 'spatial_domains'], ncols=2)
plt.savefig('clusters_comparison.pdf')

Step 6: Visualization

python
# Gene expression in space
genes = ['EPCAM', 'VIM', 'PTPRC', 'COL1A1']
sc.pl.spatial(adata, color=genes, ncols=2, spot_size=1.5, cmap='viridis')
plt.savefig('marker_genes_spatial.pdf')

# Cluster markers in space
sc.tl.rank_genes_groups(adata, 'leiden', method='wilcoxon')
sc.pl.rank_genes_groups_dotplot(adata, n_genes=5)
plt.savefig('cluster_markers.pdf')

# Save
adata.write('spatial_analyzed.h5ad')

Complete Workflow Script

python
import scanpy as sc
import squidpy as sq
import matplotlib.pyplot as plt
import os

# Configuration
data_dir = 'spaceranger_output'
output_dir = 'spatial_results'
os.makedirs(output_dir, exist_ok=True)
os.makedirs(f'{output_dir}/plots', exist_ok=True)

# Load
print('Loading data...')
adata = sq.read.visium(data_dir)
print(f'Loaded: {adata.n_obs} spots, {adata.n_vars} genes')

# QC
print('QC filtering...')
adata.var['mt'] = adata.var_names.str.startswith('MT-')
sc.pp.calculate_qc_metrics(adata, qc_vars=['mt'], inplace=True)
sc.pp.filter_cells(adata, min_counts=500)
sc.pp.filter_genes(adata, min_cells=10)
adata = adata[adata.obs.pct_counts_mt < 25, :]
print(f'After QC: {adata.n_obs} spots')

# Normalize and cluster
print('Processing...')
adata.layers['counts'] = adata.X.copy()
sc.pp.normalize_total(adata, target_sum=1e4)
sc.pp.log1p(adata)
sc.pp.highly_variable_genes(adata, n_top_genes=2000)
adata.raw = adata
adata = adata[:, adata.var.highly_variable]
sc.pp.scale(adata, max_value=10)
sc.tl.pca(adata, n_comps=50)
sc.pp.neighbors(adata, n_neighbors=15, n_pcs=30)
sc.tl.leiden(adata, resolution=0.5)

# Spatial analysis
print('Spatial analysis...')
sq.gr.spatial_neighbors(adata, coord_type='generic', n_neighs=6)
sq.gr.nhood_enrichment(adata, cluster_key='leiden')
sq.gr.spatial_autocorr(adata, mode='moran', n_perms=100)

# Plots
print('Creating plots...')
sc.pl.spatial(adata, color='leiden', spot_size=1.5, save='_clusters.pdf')
sq.pl.nhood_enrichment(adata, cluster_key='leiden', save='_nhood.pdf')

# Save
adata.write(f'{output_dir}/spatial_analyzed.h5ad')
print(f'Results saved to {output_dir}/')

Related Skills

spatial-transcriptomics/spatial-data-io - Loading formats
spatial-transcriptomics/spatial-preprocessing - QC details
spatial-transcriptomics/spatial-statistics - Moran's I, co-occurrence
spatial-transcriptomics/spatial-domains - Domain detection methods
spatial-transcriptomics/spatial-deconvolution - Cell type estimation

Illustrative Want to see what bio-workflows-spatial-pipeline would put in front of your model? Preview the rendered body and check its trigger phrases — no LLM call, no signup.

Latest benchmark · publisher runVerified

8/2/2026, 17:24 UTC · 22 cases

Lift

+27%

82% vs 55%

Agent turns

1.0 vs 1.0

Tokens

+67%

3,792.09 vs 2,274.32

fail → pass · 6pass → pass · 12fail → fail · 4

Per-case results (22)

Case	Status	Duration (ms)			Turns			Tokens			Tool calls
Case	Status	Without	With	Δ	Without	With	Δ	Without	With	Δ	Without	With	Δ
case-01 I need a Python script to process raw Visium spatial data with Scanpy and Squidpy. It should calculate QC metrics for mitochondrial content and spot counts, filter out poor-quality spots and low-expression genes, perform total count normalization and HVG selection, run PCA, neighborhood graph construction, UMAP, and Leiden clustering, then plot cluster distributions in spatial coordinates along with spatial expression maps for key tissue marker genes and save the figures.	fail→fail	24,173	16,856	-30%	1	1	0%	5,528	6,134	+11%	0	0	—
case-02 We are running domain detection on 10x Visium tissue sections using graph-based clustering. Someone suggested running standard Leiden clustering directly on transcriptomic neighbors or using Seurat spatial anchors. What is the standard approach to incorporate spatial topology into Leiden domain detection, and what specific parameters or adjacency matrices should be passed to Leiden?	fail→fail	16,480	13,936	-15%	1	1	0%	2,934	5,192	+77%	0	0	—
case-11 We are running dimensionality reduction and cell neighborhood calculation on spatial transcriptomics spots. A developer recommended 20 PCA components and 30 neighbors on 50 PCs. What exact number of PCA components and neighbor graph settings should be configured?	fail→pass	11,508	6,987	-39%	1	1	0%	1,901	3,644	+92%	0	0	—
case-12 What clustering algorithm and resolution parameter should be used for initial transcriptomic spot clustering before conducting spatial neighborhood analysis?	pass→pass	11,354	4,864	-57%	1	1	0%	1,885	3,071	+63%	0	0	—
case-22 I want to estimate cell type abundances per Visium spot using a single-cell reference matrix with cell2location in PyTorch. How do I configure and train the cell2location model?	pass→pass	16,919	14,347	-15%	1	1	0%	3,197	5,276	+65%	0	0	—
case-03 I have Space Ranger output directories containing 10x Visium tissue slides. Should I manually parse spot coordinates and image PNGs, or load matrix files directly with custom Pandas scripts? Write a Python snippet to load the Visium directory into an AnnData object.	fail→pass	9,708	5,575	-43%	1	1	0%	1,779	3,204	+80%	0	0	—
case-04 I am analyzing subcellular Xenium in situ imaging transcriptomics data output from the Xenium Onboard Analysis software. What function should I call to load this directory directly into a spatial AnnData structure?	pass→pass	7,566	3,618	-52%	1	1	0%	1,522	2,809	+85%	0	0	—
case-05 I loaded a filtered 10x H5 feature matrix using sc.read_10x_h5 for a spatial dataset. Where in the AnnData object structure must spatial metadata and image coordinates be placed so downstream spatial tools recognize it?	pass→pass	13,256	9,636	-27%	1	1	0%	2,612	4,146	+59%	0	0	—
case-06 When analyzing human Visium transcriptomics samples, how should mitochondrial genes be identified from gene symbols before calculating QC metrics, and what mitochondrial percentage ceiling should be used for filtering?	fail→pass	12,122	6,816	-44%	1	1	0%	2,202	3,415	+55%	0	0	—
case-07 In spatial Visium data preprocessing, a colleague suggests setting min_counts to 2000 and min_genes to 1000 per spot to eliminate noise. What filtering thresholds for spot counts, spot gene counts, and gene cell coverage should be applied according to standard workflow guidelines?	fail→fail	14,463	8,299	-43%	1	1	0%	2,517	3,796	+51%	0	0	—
case-08 Before performing total count normalization (10,000 target sum) and log1p transformation on a spatial transcriptomics dataset, where should the original unnormalized counts matrix be preserved inside the AnnData object?	pass→pass	6,667	4,363	-35%	1	1	0%	1,252	3,005	+140%	0	0	—
case-09 We need to select highly variable genes for downstream PCA and clustering of Visium spatial spots. How many top highly variable genes should be selected during preprocessing?	fail→pass	11,318	6,119	-46%	1	1	0%	1,994	3,377	+69%	0	0	—
case-10 Right before filtering the main AnnData object to only highly variable genes and scaling data for PCA, how should the full gene expression state be preserved for downstream marker gene visualization?	pass→pass	11,697	7,735	-34%	1	1	0%	2,172	3,718	+71%	0	0	—
case-13 When constructing a spatial graph to compute neighborhood enrichment and spatial autocorrelation, what coordinate type and neighbor count parameters should be passed to Squidpy's spatial neighbors builder?	pass→pass	13,257	6,630	-50%	1	1	0%	2,555	3,534	+38%	0	0	—
case-14 We want to test whether specific Leiden clusters preferentially neighbor each other in physical spatial layout. Write the Python calls to compute and plot neighborhood enrichment.	fail→pass	9,796	5,796	-41%	1	1	0%	1,801	3,333	+85%	0	0	—
case-15 How can we compute and plot cluster co-occurrence across spatial distances in Squidpy across Leiden clusters?	pass→pass	11,157	7,609	-32%	1	1	0%	2,266	3,568	+57%	0	0	—
case-16 We need to identify spatially variable genes across the slide using spatial autocorrelation analysis. What statistical mode and permutation count should be passed to Squidpy's spatial autocorrelation tool?	fail→pass	9,630	5,468	-43%	1	1	0%	1,878	3,298	+76%	0	0	—
case-17 After running Moran's I spatial autocorrelation on AnnData in Squidpy, in which slot inside `adata.uns` are the autocorrelation statistics stored, and by which column should results be sorted to rank top spatially variable genes?	pass→pass	6,487	3,426	-47%	1	1	0%	1,188	2,954	+149%	0	0	—
case-18 Write Python code to display spatial expression maps for tissue markers EPCAM, VIM, PTPRC, and COL1A1 side by side with continuous viridis color maps.	fail→fail	11,756	7,748	-34%	1	1	0%	2,145	3,711	+73%	0	0	—
case-19 Which statistical test and function should be executed to identify cluster marker genes across Leiden transcriptomic clusters, and which dotplot visualization function presents the top 5 markers per cluster?	pass→pass	7,072	2,969	-58%	1	1	0%	1,392	2,813	+102%	0	0	—
case-20 I have a counts matrix from a bulk RNA-seq experiment across 12 patient tissue samples with two treatment groups. How should I set up differential expression testing using DESeq2 in R to get shrinkage log2 fold changes?	pass→pass	11,740	10,385	-12%	1	1	0%	2,289	4,287	+87%	0	0	—
case-21 How do I perform peak calling and motif enrichment analysis on single-cell ATAC-seq data using ArchR in R?	pass→pass	15,195	13,834	-9%	1	1	0%	3,026	5,141	+70%	0	0	—

These cases were also screened by the skill’s intent / safety pass.See the Trust review ↗Reproduce this benchmark in Colab ↗

How this skill was measured

DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +27 percentage points is the difference between those two pass rates over the 22 comparable cases.

Model measured: gemini-3.6-flash
Graded by: gemini-3.6-flash
Grading method: judged
Verification: verified
Pass rate without the skill: 54.6%
Pass rate with the skill: 81.8%
Cases attempted: 22
Cases counted toward lift: 22
Cases that regressed: 0
Skill version measured: v2
Run completed: 8/2/2026, 17:24 UTC

Measured example — same prompt, with vs. without the skill · from the benchmark run · case case-11

Prompt

We are running dimensionality reduction and cell neighborhood calculation on spatial transcriptomics spots. A developer recommended 20 PCA components and 30 neighbors on 50 PCs. What exact number of PCA components and neighbor graph settings should be configured?

Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.

Show both answers

SkillScore blends 4 signals1 of 4 complete

BenchmarkPass rate WITH skill on a verified run82

Live evalNeeds ≥10 evaluated traces in 30 days

AI ratingNeeds ≥10 rated traces

AdoptionNeeds ≥10 serves across ≥2 orgs in 30 days

Efficiency · separate axis (at equal correctness)+0% turns+75% tokens

Shown for transparency; not credited unless a verified run shows no correctness regression.

Backed by 1 of 4 signals — each added signal makes the score more robust.

Verified on gemini-3.6-flash · re-tested 50 days ago

Model	Method	Date	Lift
gemini-3.6-flash	verified	7/24/2026	+27%

Other measured skills in the registry, with their headline benchmark lift.