Cell-Cell Interaction Analysis

Overview

This skill answers: How do cells communicate with each other in this tissue?

Cell communication fundamentally requires:

1. A sender cell expressing a ligand
2. A receiver cell expressing a receptor
3. Spatial proximity enabling the interaction

Prerequisites

Before running cell communication analysis:

• Cell type annotations available (from deconvolution or direct annotation)
• Species identified (human or mouse)
• Biological question defined (specific pathways? global patterns?)

Method Selection

Decision Framework

Q: What level of analysis do you need?
│
├─ Ligand-receptor pairs only
│   └─ LIANA+ (Python) - Multi-method consensus, recommended default
│
├─ Signaling pathway analysis
│   └─ CellChat (R) - Pathway-level insights, communication networks
│
├─ Fast computation needed
│   └─ FastCCC (C++) - Human only, very fast
│
└─ Classic/benchmark comparison
    └─ CellPhoneDB - Original method, good for comparisons

Method Comparison

Workflow

Step 1: Verify Cell Type Annotations

Required in adata.obs:
- Cell type column (e.g., 'cell_type', 'annotation')
- Clean labels (no "unknown", "unassigned")
- Biologically meaningful categories

Step 2: Configure Species

Species-specific databases:
- Human: liana_resource="consensus" (default)
- Mouse: liana_resource="mouseconsensus"

Step 3: Run Analysis

Use analyze_cell_communication tool with:

• cell_type_key: Column name for cell type annotations
• species: "human" or "mouse"
• method: Selected from decision framework

Step 4: Interpret Results

Key outputs:

• Ligand-receptor pairs: Specific molecular interactions
• Interaction scores: Strength/significance of communication
• Cell type pairs: Which cells are talking to which
• Pathways (CellChat): Grouped signaling pathways

Step 5: Visualize

Use visualize_data with plot_type="communication":

• subtype="dotplot": Overview of all interactions
• subtype="circle_plot": Network of cell type communication
• subtype="tileplot": Heatmap of interaction strength

Spatial Considerations

Adding Spatial Context

Standard cell communication ignores distance. To incorporate spatial proximity:

1. Pre-filter by spatial neighbors: Only consider cells that are spatially adjacent
2. Use spatial statistics: Correlate communication scores with co-localization

Interpreting Spatial Patterns

Questions to consider:

• Are communicating cell types spatially co-localized?
• Do interaction hotspots correspond to specific tissue regions?
• How does communication differ across spatial domains?

Common Analysis Scenarios

Scenario 1: Tumor Microenvironment

Focus on:
- Tumor-immune interactions
- CAF-tumor communication
- Immune checkpoint pathways (PD-1/PD-L1)

Scenario 2: Development/Differentiation

Focus on:
- Signaling gradients (Wnt, BMP, Notch)
- Niche-stem cell interactions
- Temporal progression of communication

Scenario 3: Tissue Architecture

Focus on:
- ECM-cell interactions
- Structural organization signals
- Regional communication patterns

Output Interpretation Guide

Significant Interactions

Look for:

• High specificity scores (interaction specific to certain cell pairs)
• Biological relevance (known pathways in tissue type)
• Spatial coherence (co-localization of interacting cells)

Common Patterns

Troubleshooting