scanpy

Name: scanpy
Author: LeonChaoX

698stars61forks

How to install scanpy?

leonchaox/qinyan-academic-skills/scanpy

$npx -y skills add LeonChaoX/qinyan-academic-skills --skill scanpy

Installs into the current project.

Summary

Standard single-cell RNA-seq analysis pipeline. Use for QC, normalization, dimensionality reduction (PCA/UMAP/t-SNE), clustering, differential expression, and visualization. Best for exploratory scRNA-seq analysis with established workflows. For deep learning models use scvi-tool

Files · 1

View on GitHub

SKILL.md

1	# Scanpy: Single-Cell Analysis
2
3	## Overview
4
5	Scanpy is a scalable Python toolkit for analyzing single-cell RNA-seq data, built on AnnData. Apply this skill for complete single-cell workflows including quality control, normalization, dimensionality reduction, clustering, marker gene identification, visualization, and trajectory analysis.
6
7	## When to Use This Skill
8
9	This skill should be used when:
10	- Analyzing single-cell RNA-seq data (.h5ad, 10X, CSV formats)
11	- Performing quality control on scRNA-seq datasets
12	- Creating UMAP, t-SNE, or PCA visualizations
13	- Identifying cell clusters and finding marker genes
14	- Annotating cell types based on gene expression
15	- Conducting trajectory inference or pseudotime analysis
16	- Generating publication-quality single-cell plots
17
18	## Quick Start
19
20	### Basic Import and Setup
21
22	```python
23	import scanpy as sc
24	import pandas as pd
25	import numpy as np
26
27	# Configure settings
28	sc.settings.verbosity = 3
29	sc.settings.set_figure_params(dpi=80, facecolor='white')
30	sc.settings.figdir = './figures/'
31	```
32
33	### Loading Data
34
35	```python
36	# From 10X Genomics
37	adata = sc.read_10x_mtx('path/to/data/')
38	adata = sc.read_10x_h5('path/to/data.h5')
39
40	# From h5ad (AnnData format)
41	adata = sc.read_h5ad('path/to/data.h5ad')
42
43	# From CSV
44	adata = sc.read_csv('path/to/data.csv')
45	```
46
47	### Understanding AnnData Structure
48
49	The AnnData object is the core data structure in scanpy:
50
51	```python
52	adata.X # Expression matrix (cells × genes)
53	adata.obs # Cell metadata (DataFrame)
54	adata.var # Gene metadata (DataFrame)
55	adata.uns # Unstructured annotations (dict)
56	adata.obsm # Multi-dimensional cell data (PCA, UMAP)
57	adata.raw # Raw data backup
58
59	# Access cell and gene names
60	adata.obs_names # Cell barcodes
61	adata.var_names # Gene names
62	```
63
64	## Standard Analysis Workflow
65
66	### 1. Quality Control
67
68	Identify and filter low-quality cells and genes:
69
70	```python
71	# Identify mitochondrial genes
72	adata.var['mt'] = adata.var_names.str.startswith('MT-')
73
74	# Calculate QC metrics
75	sc.pp.calculate_qc_metrics(adata, qc_vars=['mt'], inplace=True)
76
77	# Visualize QC metrics
78	sc.pl.violin(adata, ['n_genes_by_counts', 'total_counts', 'pct_counts_mt'],
79	jitter=0.4, multi_panel=True)
80
81	# Filter cells and genes
82	sc.pp.filter_cells(adata, min_genes=200)
83	sc.pp.filter_genes(adata, min_cells=3)
84	adata = adata[adata.obs.pct_counts_mt < 5, :] # Remove high MT% cells
85	```
86
87	Use the QC script for automated analysis:
88	```bash
89	python scripts/qc_analysis.py input_file.h5ad --output filtered.h5ad
90	```
91
92	### 2. Normalization and Preprocessing
93
94	```python
95	# Normalize to 10,000 counts per cell
96	sc.pp.normalize_total(adata, target_sum=1e4)
97
98	# Log-transform
99	sc.pp.log1p(adata)
100
101	# Save raw counts for later
102	adata.raw = adata
103
104	# Identify highly variable genes
105	sc.pp.highly_variable_genes(adata, n_top_genes=2000)
106	sc.pl.highly_variable_genes(adata)
107
108	# Subset to highly variable genes
109	adata = adata[:, adata.var.highly_variable]
110
111	# Regress out unwanted variation
112	sc.pp.regress_out(adata, ['total_counts', 'pct_counts_mt'])
113
114	# Scale data
115	sc.pp.scale(adata, max_value=10)
116	```
117
118	### 3. Dimensionality Reduction
119
120	```python
121	# PCA
122	sc.tl.pca(adata, svd_solver='arpack')
123	sc.pl.pca_variance_ratio(adata, log=True) # Check elbow plot
124
125	# Compute neighborhood graph
126	sc.pp.neighbors(adata, n_neighbors=10, n_pcs=40)
127
128	# UMAP for visualization
129	sc.tl.umap(adata)
130	sc.pl.umap(adata, color='leiden')
131
132	# Alternative: t-SNE
133	sc.tl.tsne(adata)
134	```
135
136	### 4. Clustering
137
138	```python
139	# Leiden clustering (recommended)
140	sc.tl.leiden(adata, resolution=0.5)
141	sc.pl.umap(adata, color='leiden', legend_loc='on data')
142
143	# Try multiple resolutions to find optimal granularity
144	for res in [0.3, 0.5, 0.8, 1.0]:
145	sc.tl.leiden(adata, resolution=res, key_added=f'leiden_{res}')
146	```
147
148	### 5. Marker Gene Identification
149
150	```python
151	# Find marker genes for each cluster
152	sc.tl.rank_genes_groups(adata, 'leiden', method='wilcoxon')
153
154	# Visualize results
155	sc.pl.rank_genes_groups(adata, n_genes=25, sharey=False)
156	sc.pl.rank_genes_groups_heatmap(adata, n_genes=10)
157	sc.pl.rank_genes_groups_dotplot(adata, n_genes=5)
158
159	# Get results as DataFrame
160	markers = sc.get.rank_genes_groups_df(adata, group='0')
161	```
162
163	### 6. Cell Type Annotation
164
165	```python
166	# Define marker genes for known cell types
167	marker_genes = ['CD3D', 'CD14', 'MS4A1', 'NKG7', 'FCGR3A']
168
169	# Visualize markers
170	sc.pl.umap(adata, color=marker_genes, use_raw=True)
171	sc.pl.dotplot(adata, var_names=marker_genes, groupby='leiden')
172
173	# Manual annotation
174	cluster_to_celltype = {
175	'0': 'CD4 T cells',
176	'1': 'CD14+ Monocytes',
177	'2': 'B cells',
178	'3': 'CD8 T ce

People who liked this also viewed

6 picks

Type

scanpy

698stars61forks

View on GitHub

How to install scanpy?

leonchaox/qinyan-academic-skills/scanpy

$npx -y skills add LeonChaoX/qinyan-academic-skills --skill scanpy

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Scanpy: Single-Cell Analysis
2
3	## Overview
4
5	Scanpy is a scalable Python toolkit for analyzing single-cell RNA-seq data, built on AnnData. Apply this skill for complete single-cell workflows including quality control, normalization, dimensionality reduction, clustering, marker gene identification, visualization, and trajectory analysis.
6
7	## When to Use This Skill
8
9	This skill should be used when:
10	- Analyzing single-cell RNA-seq data (.h5ad, 10X, CSV formats)
11	- Performing quality control on scRNA-seq datasets
12	- Creating UMAP, t-SNE, or PCA visualizations
13	- Identifying cell clusters and finding marker genes
14	- Annotating cell types based on gene expression
15	- Conducting trajectory inference or pseudotime analysis
16	- Generating publication-quality single-cell plots
17
18	## Quick Start
19
20	### Basic Import and Setup
21
22	```python
23	import scanpy as sc
24	import pandas as pd
25	import numpy as np
26
27	# Configure settings
28	sc.settings.verbosity = 3
29	sc.settings.set_figure_params(dpi=80, facecolor='white')
30	sc.settings.figdir = './figures/'
31	```
32
33	### Loading Data
34
35	```python
36	# From 10X Genomics
37	adata = sc.read_10x_mtx('path/to/data/')
38	adata = sc.read_10x_h5('path/to/data.h5')
39
40	# From h5ad (AnnData format)
41	adata = sc.read_h5ad('path/to/data.h5ad')
42
43	# From CSV
44	adata = sc.read_csv('path/to/data.csv')
45	```
46
47	### Understanding AnnData Structure
48
49	The AnnData object is the core data structure in scanpy:
50
51	```python
52	adata.X # Expression matrix (cells × genes)
53	adata.obs # Cell metadata (DataFrame)
54	adata.var # Gene metadata (DataFrame)
55	adata.uns # Unstructured annotations (dict)
56	adata.obsm # Multi-dimensional cell data (PCA, UMAP)
57	adata.raw # Raw data backup
58
59	# Access cell and gene names
60	adata.obs_names # Cell barcodes
61	adata.var_names # Gene names
62	```
63
64	## Standard Analysis Workflow
65
66	### 1. Quality Control
67
68	Identify and filter low-quality cells and genes:
69
70	```python
71	# Identify mitochondrial genes
72	adata.var['mt'] = adata.var_names.str.startswith('MT-')
73
74	# Calculate QC metrics
75	sc.pp.calculate_qc_metrics(adata, qc_vars=['mt'], inplace=True)
76
77	# Visualize QC metrics
78	sc.pl.violin(adata, ['n_genes_by_counts', 'total_counts', 'pct_counts_mt'],
79	jitter=0.4, multi_panel=True)
80
81	# Filter cells and genes
82	sc.pp.filter_cells(adata, min_genes=200)
83	sc.pp.filter_genes(adata, min_cells=3)
84	adata = adata[adata.obs.pct_counts_mt < 5, :] # Remove high MT% cells
85	```
86
87	Use the QC script for automated analysis:
88	```bash
89	python scripts/qc_analysis.py input_file.h5ad --output filtered.h5ad
90	```
91
92	### 2. Normalization and Preprocessing
93
94	```python
95	# Normalize to 10,000 counts per cell
96	sc.pp.normalize_total(adata, target_sum=1e4)
97
98	# Log-transform
99	sc.pp.log1p(adata)
100
101	# Save raw counts for later
102	adata.raw = adata
103
104	# Identify highly variable genes
105	sc.pp.highly_variable_genes(adata, n_top_genes=2000)
106	sc.pl.highly_variable_genes(adata)
107
108	# Subset to highly variable genes
109	adata = adata[:, adata.var.highly_variable]
110
111	# Regress out unwanted variation
112	sc.pp.regress_out(adata, ['total_counts', 'pct_counts_mt'])
113
114	# Scale data
115	sc.pp.scale(adata, max_value=10)
116	```
117
118	### 3. Dimensionality Reduction
119
120	```python
121	# PCA
122	sc.tl.pca(adata, svd_solver='arpack')
123	sc.pl.pca_variance_ratio(adata, log=True) # Check elbow plot
124
125	# Compute neighborhood graph
126	sc.pp.neighbors(adata, n_neighbors=10, n_pcs=40)
127
128	# UMAP for visualization
129	sc.tl.umap(adata)
130	sc.pl.umap(adata, color='leiden')
131
132	# Alternative: t-SNE
133	sc.tl.tsne(adata)
134	```
135
136	### 4. Clustering
137
138	```python
139	# Leiden clustering (recommended)
140	sc.tl.leiden(adata, resolution=0.5)
141	sc.pl.umap(adata, color='leiden', legend_loc='on data')
142
143	# Try multiple resolutions to find optimal granularity
144	for res in [0.3, 0.5, 0.8, 1.0]:
145	sc.tl.leiden(adata, resolution=res, key_added=f'leiden_{res}')
146	```
147
148	### 5. Marker Gene Identification
149
150	```python
151	# Find marker genes for each cluster
152	sc.tl.rank_genes_groups(adata, 'leiden', method='wilcoxon')
153
154	# Visualize results
155	sc.pl.rank_genes_groups(adata, n_genes=25, sharey=False)
156	sc.pl.rank_genes_groups_heatmap(adata, n_genes=10)
157	sc.pl.rank_genes_groups_dotplot(adata, n_genes=5)
158
159	# Get results as DataFrame
160	markers = sc.get.rank_genes_groups_df(adata, group='0')
161	```
162
163	### 6. Cell Type Annotation
164
165	```python
166	# Define marker genes for known cell types
167	marker_genes = ['CD3D', 'CD14', 'MS4A1', 'NKG7', 'FCGR3A']
168
169	# Visualize markers
170	sc.pl.umap(adata, color=marker_genes, use_raw=True)
171	sc.pl.dotplot(adata, var_names=marker_genes, groupby='leiden')
172
173	# Manual annotation
174	cluster_to_celltype = {
175	'0': 'CD4 T cells',
176	'1': 'CD14+ Monocytes',
177	'2': 'B cells',
178	'3': 'CD8 T ce

People who liked this also viewed

6 picks

Type