cellxgene

cellxgene: Interactive Single-Cell Data Explorer

Overview

cellxgene is an interactive web-based tool for exploring single-cell RNA-seq datasets. It provides a fast, intuitive interface for visualizing cell populations, gene expression patterns, and metadata without requiring programming knowledge.

When to Use This Skill

This skill should be used when:

• Exploring single-cell datasets interactively
• Creating shareable visualizations for collaborators
• Checking gene expression across cell types
• Filtering and subsetting data visually
• Creating annotations and cell labels
• Preparing figures for publications
• Sharing datasets with collaborators or publicly

Quick Start

Installation

# Install via pip
pip install cellxgene

# Or using conda
conda install -c conda-forge cellxgene

Launch cellxgene

# Launch with a h5ad file
cellxgene launch data.h5ad

# Launch with specific host and port
cellxgene launch data.h5ad --port 5000 --host 0.0.0.0

# Launch with annotations
cellxgene launch data.h5ad --annotations ./annotations.tsv

Opening Data

1. Navigate to http://localhost:5000 (or your specified port)
2. The interface loads automatically with the dataset

Interface Overview

Main View Components

• Left sidebar: Gene expression search, metadata filters
• Center: Scatter plot (UMAP, t-SNE, PCA)
• Bottom: Gene expression violin/dot plots
• Right panel: Categorical color by options

Navigation Controls

• Scroll: Zoom in/out
• Click + drag: Pan
• Shift + click: Select points
• Double-click: Reset view

Common Tasks

Searching Genes

1. Type gene name in the search box (top left)
2. Gene expression automatically visualizes on the plot
3. Use the "color by" dropdown to switch visualization

Filtering Data

# Create a filtered dataset for cellxgene
import scanpy as sc

adata = sc.read_h5ad("data.h5ad")

# Filter to specific cell types
adata_filtered = adata[adata.obs['cell_type'].isin(['T cells', 'B cells'])]

# Save for cellxgene
adata_filtered.write_h5ad("filtered_data.h5ad")