Skip to contents

Connectome Connectome logo

Computational Framework for Cell-Cell Communication Analysis in Single-Cell Transcriptomics

📖 Full Documentation | 🚀 Get Started | 📊 Visualization Gallery

Overview

Connectome is an R package designed for systematic inference and visualization of cell-cell communication networks from single-cell RNA sequencing (scRNA-seq) data. The package constructs connectomic edgelists representing intercellular signaling topologies based on ligand-receptor co-expression patterns across distinct cell populations.

The analytical framework integrates:

  • FANTOM5 ligand-receptor database with multi-species support (human, mouse, rat, pig)
  • Statistical significance testing via Wilcoxon rank-sum tests
  • Network centrality analysis using Kleinberg hub and authority scores
  • Differential connectivity analysis for comparative studies

Citation

If you use Connectome in your research, please cite:

Raredon, M.S.B., Yang, J., Garritano, J. et al. Computation and visualization of cell–cell signaling topologies in single-cell systems data using Connectome. Scientific Reports 12, 4187 (2022). https://doi.org/10.1038/s41598-022-07959-x

BibTeX:

@article{raredon2022connectome,
  title={Computation and visualization of cell--cell signaling topologies in single-cell systems data using Connectome},
  author={Raredon, Micha Sam Brickman and Yang, Junchen and Garritano, James and others},
  journal={Scientific Reports},
  volume={12},
  pages={4187},
  year={2022},
  publisher={Nature Publishing Group},
  doi={10.1038/s41598-022-07959-x}
}

Installation 

install.packages("Connectome", repos = "https://zaoqu-liu.r-universe.dev")

From GitHub 

if (!requireNamespace("remotes", quietly = TRUE))
    install.packages("remotes")
remotes::install_github("Zaoqu-Liu/Connectome")

Dependencies

Required Bioconductor packages:

if (!requireNamespace("BiocManager", quietly = TRUE))
    install.packages("BiocManager")
BiocManager::install(c("circlize", "ComplexHeatmap"))

Quick Start 

library(Seurat)
library(Connectome)

# Ensure data is normalized and scaled
seurat_obj <- NormalizeData(seurat_obj)
seurat_obj <- ScaleData(seurat_obj)

# Construct connectome
connectome <- CreateConnectome(
  object = seurat_obj,
  species = "human",
  LR.database = "fantom5",
  min.cells.per.ident = 75,
  p.values = TRUE
)

# Apply biological filters
connectome_filtered <- FilterConnectome(
  connectome,
  min.pct = 0.1,      # Minimum expression fraction

  min.z = 0.25,       # Minimum z-score threshold
  max.p = 0.05        # Maximum adjusted p-value
)

# Visualize network topology
NetworkPlot(connectome_filtered)
CircosPlot(connectome_filtered)

Core Functions

Network Construction

Function Description
CreateConnectome() Construct connectomic edgelist from Seurat object
FilterConnectome() Apply expression, significance, and topological filters
DifferentialConnectome() Compute differential connectivity between conditions
SingleCellConnectome() Generate single-cell resolution connectivity matrix

Visualization

Function Description
NetworkPlot() igraph-based directed network visualization
CircosPlot() Chord diagram representation of connectivity
CircosDiff() Differential connectivity chord diagram
EdgeDotPlot() Dot plot matrix of edge weights
DiffEdgeDotPlot() Differential edge visualization

Network Analysis

Function Description
Centrality() Hub and authority score analysis by signaling mode
CompareCentrality() Cross-condition centrality comparison
ModalDotPlot() Mode-stratified centrality visualization
SignalScatter() Ligand-receptor co-expression scatter plot

Methodological Framework

Edge Weight Computation

Connectome computes edge weights using ligand and receptor expression values:

wijLR=f(EiL,EjR)w_{ij}^{LR} = f(E_i^L, E_j^R)

where EiLE_i^L denotes ligand expression in source population ii, EjRE_j^R denotes receptor expression in target population jj, and ff can be configured as:

  • Product: w=EL×ERw = E^L \times E^R (default, captures co-expression)
  • Sum: w=EL+ERw = E^L + E^R (additive contribution)
  • Mean: w=(EL+ER)/2w = (E^L + E^R) / 2 (average signal strength)

Statistical Testing

Significance is assessed via Wilcoxon rank-sum tests comparing gene expression between each cluster and all other cells, with Bonferroni correction for multiple testing.

Diagnostic Odds Ratio (DOR)

Gene specificity is quantified using the log-transformed Diagnostic Odds Ratio with Haldane-Anscombe correction:

DOR=log⁡((TP+0.5)(TN+0.5)(FP+0.5)(FN+0.5))\text{DOR} = \log\left(\frac{(TP + 0.5)(TN + 0.5)}{(FP + 0.5)(FN + 0.5)}\right)

Supported Species

The package includes curated ligand-receptor databases from FANTOM5 for:

Species Dataset Pairs
Human ncomms8866_human 2,557
Mouse ncomms8866_mouse ~2,400
Rat ncomms8866_rat ~2,300
Pig ncomms8866_pig ~2,200

Custom ligand-receptor databases can be supplied via the custom.list parameter.

Integration with LIANA

Connectome is integrated with LIANA (Ligand-Receptor Analysis framework):

library(liana)
liana_res <- liana_wrap(sce, method = "connectome")

System Requirements

  • R version: ≥ 4.0.0
  • Core dependencies: Seurat (≥ 4.0.0), igraph, circlize, ComplexHeatmap
  • Memory: Scales with dataset size; recommend 8+ GB RAM for large datasets
  • Platform: Cross-platform (macOS, Linux, Windows)

License

GPL-3.0 © 2022 Micha Sam Brickman Raredon, Zaoqu Liu

References

  1. Raredon, M.S.B., Yang, J., Garritano, J. et al. Computation and visualization of cell–cell signaling topologies in single-cell systems data using Connectome. Sci Rep 12, 4187 (2022). DOI: 10.1038/s41598-022-07959-x

  2. Ramilowski, J.A. et al. A draft network of ligand–receptor-mediated multicellular signalling in human. Nat Commun 6, 7866 (2015). DOI: 10.1038/ncomms8866

  3. Raredon, M.S.B. et al. Single-cell connectomic analysis of adult mammalian lungs. Science Advances 5(12), eaaw3851 (2019). DOI: 10.1126/sciadv.aaw3851

Contact