tools-only Cell-Cell Communication - Usage Guide
Overview
Cell-cell communication analysis identifies ligand-receptor interactions between cell types from scRNA-seq data to understand tissue organization, signaling networks, and intercellular coordination.
Prerequisites
# CellChat
devtools::install_github('sqjin/CellChat')
# NicheNet
devtools::install_github('saeyslab/nichenetr')
# Download NicheNet databases from Zenodo
# LIANA (Python)
pip install liana
Quick Start
Tell your AI agent what you want to do:
- "Find ligand-receptor interactions between cell types"
- "Compare cell communication between conditions"
- "Identify signaling pathways between macrophages and T cells"
Example Prompts
Interaction Discovery
"Run CellChat to find all ligand-receptor interactions" "Identify which cell types communicate the most" "Find interactions specific to fibroblast-epithelial crosstalk"
Pathway Analysis
"Which signaling pathways are active between these cell types?" "Show a network of WNT signaling interactions" "Find all TGF-beta pathway communications"
Condition Comparison
"Compare cell communication between control and disease" "Which interactions are gained or lost in treatment?" "Show differential signaling between conditions"
Target Prediction
"Use NicheNet to find ligands affecting gene expression in T cells" "Which ligands from macrophages activate inflammatory genes?" "Predict downstream targets of fibroblast-derived signals"
Visualization
"Create a chord diagram of cell communications" "Show a heatmap of interaction strengths" "Plot the ligand-receptor network"
What the Agent Will Do
- Load annotated scRNA-seq data with cell types
- Select appropriate ligand-receptor database
- Calculate communication probabilities
- Identify significant interactions
- Aggregate at pathway level
- Visualize networks and interactions
- Compare conditions if applicable
Tool Selection
| Tool | Strengths | Language | Best For |
|---|---|---|---|
| CellChat | Comprehensive, pathway-level | R | General CCC, comparison |
| NicheNet | Ligand-target prediction | R | Functional impact on receivers |
| LIANA | Multiple methods, consensus | Python | Robust rankings, multi-sample |
Tips
- Cell type annotation quality directly affects results
- Expression thresholds filter noise but may miss real interactions
- Database choice affects coverage (CellChatDB vs OmniPath vs custom)
- Multiple testing - adjust p-values for many comparisons
- Validate top hits - check ligand and receptor expression patterns
- Spatial context helps - pair with spatial data if available