NOVA

📚 Documentation: https://zaoqu-liu.github.io/NOVA/

Network Of Versatile Cell-Cell Communication Analysis

NOVA is a high-performance R package for inferring and visualizing cell-to-cell communication networks from single-cell and bulk transcriptomic data. By leveraging the connectomeDB2020 ligand-receptor interaction database, NOVA enables systematic identification of intercellular signaling interactions across diverse biological contexts.

Overview

Cell-to-cell communication is fundamental to multicellular organisms, orchestrating processes from development to immune responses. NOVA provides a comprehensive computational framework for:

Key Features

Feature Description
Seurat Integration Native compatibility with Seurat V4 and V5 objects
Multi-Species Support for 21 species via NCBI HomoloGene
High Performance Vectorized operations with optional C++ acceleration
Rich Visualization Heatmaps, network graphs, chord diagrams, and more
Differential Analysis Compare communication patterns between conditions

Installation

From R-universe (Recommended)

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

From GitHub

# install.packages("remotes")
remotes::install_github("Zaoqu-Liu/NOVA")

Quick Start

Basic Usage

library(NOVA)

# From Seurat object
result <- ExtractEdges(
  seurat_obj,
  species = "mouse",
  cluster_col = "cell_type"
)

# View results
print(result)
summary(result)

Visualization

# Communication heatmap
PlotHeatmap(result, metric = "specificity")

# Network graph
PlotNetwork(result, layout = "circle")

# Chord diagram
PlotChord(result)

# LR pairs between clusters
PlotLRPairs(result, sending = "T_cells", target = "Macrophages")

Differential Analysis

# Compare two conditions
diff <- DiffEdges(ctrl_result, treat_result, log2fc_threshold = 1)

# Visualize changes
PlotDiffHeatmap(diff)
PlotVolcano(diff)

Methodology

Communication Score Calculation

NOVA computes cell-cell communication edges based on ligand-receptor co-expression:

  1. Detection Rate: Proportion of cells expressing each gene

    pct = n_expressing / n_total

  2. Expression Specificity: Cluster-level enrichment score

    specificity = mean_cluster / Σ(mean_all_clusters)

  3. Edge Weight: Product of ligand and receptor metrics

    weight = ligand_expression × receptor_expression
edge_specificity = ligand_specificity × receptor_specificity

Ligand-Receptor Database

NOVA uses connectomeDB2020, comprising:

Supported Species

NOVA supports 21 species through NCBI HomoloGene:

supported_species()
#> human, mouse, rat, zebrafish, fruitfly, chimpanzee, dog, 
#> monkey, cattle, chicken, frog, mosquito, nematode, ...

Main Functions

Function Description
ExtractEdges() Infer ligand-receptor mediated interactions
FilterEdges() Filter edges by expression/specificity thresholds
GetEdges() Retrieve edges for specific cluster pairs
DiffEdges() Differential communication analysis
PlotHeatmap() Communication strength heatmap
PlotNetwork() Network visualization
PlotChord() Chord diagram
PlotLRPairs() Bipartite L-R visualization
PlotDiffHeatmap() Differential heatmap
PlotVolcano() Volcano plot for differential analysis

Citation

If you use NOVA in your research, please cite:

@software{nova2026,
  author = {Liu, Zaoqu},
  title = {NOVA: Network Of Versatile Cell-Cell Communication Analysis},
  year = {2026},
  url = {https://github.com/Zaoqu-Liu/NOVA}
}

NOVA builds upon the connectomeDB2020 database:

Hou, R., Denisenko, E., Ong, H.T. et al. Predicting cell-to-cell communication networks using NATMI. Nat Commun 11, 5011 (2020). https://doi.org/10.1038/s41467-020-18873-z

License

This project is licensed under the MIT License - see the LICENSE file for details.

Contact

Acknowledgments