ExpressionCellNet: Network-Based Analysis of Gene Expression Perturbations

Description

Network-centric framework for integrative analysis of high-throughput gene expression data using user-supplied gene-gene interaction graphs. Constructs seed-centered multi-generation networks constrained by expression correlations and simulates expression perturbation scenarios via regression-based prediction (van Dam et al. 2018) doi:10.1093/bib/bbw139.

Author(s)

Maintainer: Ehsan Keramati ehsan.keramati14@gmail.com

Analyze network topology and identify hub genes

Description

Computes degree, closeness, betweenness, and eigenvector centrality and stores: (i) Network Analysis table and (ii) Hub Genes table (top 10

Usage

AnalyzeNetwork(ExpCellNetObj)

Arguments

Value

Updated ExpressionCellNet object with "Network Analysis" and "Hub Genes".

Examples

data("example_data")
obj <- AnalyzeNetwork(example_data$expcellnetobj)

Construct a gene interaction network centered on a seed gene

Description

Builds a multi-generation network starting from a seed gene using curated interactions and correlation filtering across generations.

Usage

BuildNetwork(
  ExpCellNetObj,
  Gene = "Gene1",
  Generation = 3,
  CorrelationThreshold = c(0.7, 0.7, 0.7)
)

Arguments

Value

Updated ExpressionCellNet object with added components such as "Network" and "Generations".

Examples

data("example_data")
obj <- BuildNetwork(
    example_data$expcellnetobj,
    Gene = "Gene1",
    Generation = 3,
    CorrelationThreshold = c(0.65, 0.70, 0.75)
)

Normalize expression count matrix

Description

Filters low-expression genes (optional), applies CPM normalization (optional), and log2 transformation (optional). Stores results inside the ExpressionCellNet object.

Usage

CountMatrixNormalization(
  ExpCellNetObj,
  CheckExpression = TRUE,
  CPM = TRUE,
  LogTransform = FALSE
)

Arguments

Value

Updated ExpressionCellNet object containing normalized matrix.

Examples

data("example_data")
obj <- CountMatrixNormalization(
    example_data$expcellnetobj,
    CheckExpression = TRUE,
    CPM = TRUE,
    LogTransform = TRUE
)

ExpressionCellNet: Network-centric modeling of gene expression perturbations

Description

ExpressionCellNet provides an integrative workflow for constructing biologically constrained gene–gene interaction networks by combining curated interaction evidence with statistically supported expression correlations. Built networks can be analyzed using centrality metrics, hub-gene identification, shortest-path queries, and enrichment analysis. The framework also supports regression-based prediction to simulate in silico perturbations (e.g., gene silencing/overexpression) and to estimate how expression changes propagate through connected genes.

Core workflow

1. Create object: createExpCellNetObj()

2. Normalize/filter: CountMatrixNormalization()

3. Build network: BuildNetwork()

4. Network analysis: AnalyzeNetwork(), ShowHubGenes(), FindPathway()

5. Prediction and visualization: NetworkPrediction(), PlotNetworkPrediction()

6. Dimension reduction: PCAforGenes(), UMAPforGenes()

Main functions

createExpCellNetObj

Create the main analysis object.

CountMatrixNormalization

Filter and normalize count matrices (CPM/log2).

BuildNetwork

Construct a multi-generation network from a seed gene.

AnalyzeNetwork

Compute centrality metrics and extract hub genes (top 10%).

FindPathway

Shortest path between two genes in the constructed network.

NetworkPrediction

Predict expression response to gene perturbation.

Input data

Required inputs typically include:

• A gene-by-sample expression matrix (counts or expression values)

• A gene annotation table with consistent gene identifiers

• A curated gene gene interaction database (two-column edge list)

Notes

This package is intended for decision support and hypothesis generation and does not claim causal inference.

Author(s)

Maintainer: Ehsan Keramati ehsan.keramati14@gmail.com

Extract network-derived expression features

Description

Extracts a feature matrix from the normalized expression data by restricting genes to those present in the constructed interaction network. The resulting matrix is intended for downstream modeling and dimensionality reduction.

Usage

ExtractFeature(ExpCellNetObj)

Arguments

Value

A data.frame or matrix of network-derived expression features (genes × samples), suitable for regression or multivariate analysis.

Examples

data("example_data")
features <- ExtractFeature(example_data$expcellnetobj)

Find the shortest path between two genes in the network

Description

Computes the shortest interaction path (as a sequence of gene names) between a start gene and an end gene using the network stored in the ExpressionCellNet object.

Usage

FindPathway(ExpCellNetObj, Start_gene, End_gene)

Arguments

Value

A character vector of gene symbols representing the shortest path.

Examples

data("example_data")
path <- FindPathway(example_data$expcellnetobj, Start_gene="Gene1", End_gene="Gene11")
path

Compute mean expression of a gene

Description

Computes mean expression from raw or normalized matrix (default: normalized).

Usage

MeanExpression(ExpCellNetObj, Gene, Normalized = TRUE)

Arguments

Value

Numeric mean expression value.

Examples

data("example_data")
MeanExpression(example_data$expcellnetobj , "Gene1")

Predict network-wide expression changes after perturbing a seed gene

Description

Fits linear regression models for seed-to-network genes and predicts shifts under a user-defined perturbation scenario. Stores multiple result components in the object.

Usage

NetworkPrediction(ExpCellNetObj, Gene, Expression)

Arguments

Value

Updated ExpressionCellNet object with regression models and results.

Examples

data("example_data")
obj <- NetworkPrediction(example_data$expcellnetobj, Gene="Gene1", Expression=0)

PCA on network-derived features

Description

Performs principal component analysis (PCA) on the network-restricted feature matrix returned by ExtractFeature(), after min-max scaling, and returns a ggplot of PC1 vs PC2.

Usage

PCAforGenes(ExpCellNetObj)

Arguments

Value

A ggplot object showing PCA embedding of genes (nodes).

Examples

data("example_data")
p <- PCAforGenes(example_data$expcellnetobj)
p

Plot correlation between two genes

Description

Plots correlation/regression between two genes (e.g., seed and distal gene).

Usage

PlotCorrelation(ExpCellNetObj, Gene1, Gene2)

Arguments

Value

Invisibly returns NULL (plot is produced).

Examples

data("example_data")
PlotCorrelation(example_data$expcellnetobj, "Gene1", "Gene11")

Plot expression matrix distribution

Description

Visualizes the distribution of raw or normalized expression values across samples.

Usage

PlotCountMatrix(ExpCellNetObj, Matrix = "Normalized")

Arguments

Value

A plot (base R) is produced as a side effect.

Examples

data("example_data")
PlotCountMatrix(example_data$expcellnetobj, Matrix = "RAW")
PlotCountMatrix(example_data$expcellnetobj, Matrix = "Normalized")

Plot network prediction summary

Description

Creates a summary plot of predicted percent changes across network genes (bar plot).

Usage

PlotNetworkPrediction(ExpCellNetObj)

Arguments

Value

Invisibly returns NULL (plot is produced).

Examples

data("example_data")
PlotNetworkPrediction(example_data$expcellnetobj)

Plot pathway-restricted prediction to a target gene

Description

Summarizes modeled expression changes along the seed-to-target pathway.

Usage

PlotPathwayPrediction(ExpCellNetObj, target_gene)

Arguments

Value

Invisibly returns NULL (plot is produced).

Examples

data("example_data")
obj <- NetworkPrediction(example_data$expcellnetobj, Gene="Gene1", Expression=0)
PlotPathwayPrediction(obj, "Gene11")

Regenerate train/test split

Description

Re-creates the internal random 80/20 train-test split without stratifimessageion.

Usage

RegenerateTrainTest(ExpCellNetObj)

Arguments

Value

Updated ExpressionCellNet object with regenerated split.

Examples

data("example_data")
obj <- RegenerateTrainTest(example_data$expcellnetobj)

Show hub genes on the network

Description

Highlights hub genes on the igraph network visualization.

Usage

ShowHubGenes(ExpCellNetObj)

Arguments

Value

Invisibly returns NULL (plot is produced).

Examples

data("example_data")
ShowHubGenes(example_data$expcellnetobj)

Visualize the constructed network

Description

Plots the current network stored in the ExpressionCellNet object.

Usage

ShowNetwork(ExpCellNetObj)

Arguments

Value

A plot (igraph/base) is produced as a side effect.

Examples

data("example_data")
ShowNetwork(example_data$expcellnetobj)

Visualize a pathway (highlighted subgraph)

Description

Highlights the nodes/edges of a provided pathway on the full network visualization.

Usage

ShowPathway(ExpCellNetObj, Pathway)

Arguments

Value

A plot (igraph/base) is produced as a side effect.

Examples

data("example_data")
p <- FindPathway(example_data$expcellnetobj, Start_gene="Gene1", End_gene="Gene11")
ShowPathway(example_data$expcellnetobj, Pathway = p)

UMAP on network-derived features

Description

Computes a 2D UMAP embedding of the network-restricted feature matrix returned by ExtractFeature() (after min-max scaling) and returns a ggplot visualization.

Usage

UMAPforGenes(ExpCellNetObj)

Arguments

Value

A ggplot object showing UMAP embedding of genes (nodes).

Examples

data("example_data")
u <- UMAPforGenes(example_data$expcellnetobj)
u

Create an ExpressionCellNet object

Description

Wraps the core inputs (count matrix, annotation table, and interaction database) into a single ExpressionCellNet object used by downstream functions.

Usage

createExpCellNetObj(count_matrix, annotations, intraction_db)

Arguments

Value

An ExpressionCellNet object (list) containing core data components.

Examples

data("example_data")
obj <- createExpCellNetObj(
   count_matrix  = example_data$counts,
   annotations   = example_data$annot_df,
   intraction_db = example_data$interactions
)

Example dataset for ExpressionCellNet

Description

Simulated example dataset containing four objects: counts, annot_df, interactions, and expcellnetobj. All data are mock / artificial for testing.

Usage

data(example_data)

Format

An RData file containing:

counts

Numeric matrix of simulated RNA-seq counts.

annot_df

Data frame with ENSEMBL IDs and Gene Symbols.

interactions

Data frame representing gene–gene interactions.

expcellnetobj

Example ExpressionCellNet object.