Chapter 19: Interactive Visualization of WGCNA Results

📥 Download Lecture Slides (PPTX)

19.1 Visualizing WGCNA Results

While static plots are useful, WGCNA generates a wealth of data that is best explored interactively. In this chapter, we build a Shiny dashboard to explore the network analysis results from Chapter 18.

19.2 Dashboard Structure

Try the interactive module-trait heatmap demo here: Interactive WGCNA Heatmap

We will use shinydashboard to organize the results into three main tabs: 1. Network Topology: Visualizing the soft thresholding and scale-free fit. 2. Module Detection: Exploring the dendrogram and module assignment. 3. Module-Trait Relationships: An interactive heatmap correlating modules with clinical variables.

19.3 Bioinformatics in Action: The Shiny App

Here is the structure of the app.R file designed to visualize WGCNA outputs.

library(shiny)
library(shinydashboard)
library(WGCNA)
library(gplots)

# 1. UI Definition
ui <- dashboardPage(
  dashboardHeader(title = "WGCNA Results Viewer"),

  dashboardSidebar(
    sidebarMenu(
      menuItem("Network Topology", tabName = "topology", icon = icon("chart-line")),
      menuItem("Dendrogram", tabName = "dendro", icon = icon("tree")),
      menuItem("Module-Trait Heatmap", tabName = "heatmap", icon = icon("th"))
    )
  ),

  dashboardBody(
    tabItems(
      # Tab 1: Topology
      tabItem(tabName = "topology",
              h2("Scale-Free Topology Fit"),
              plotOutput("sftPlot")),

      # Tab 2: Dendrogram
      tabItem(tabName = "dendro",
              h2("Clustering Dendrogram"),
              plotOutput("dendroPlot", height = "600px")),

      # Tab 3: Heatmap
      tabItem(tabName = "heatmap",
              h2("Module-Trait Relationships"),
              plotOutput("heatmapPlot", height = "800px"))
    )
  )
)

# 2. Server Logic
server <- function(input, output) {

  # In a real deployment, load the pre-computed RData from the pipeline
  # load("wgcna_results.RData")

  output$sftPlot <- renderPlot({
    # Placeholder for the Scale-Free Topology Plot generated in Step 2
    plot(1:10, type="n", xlab="Soft Threshold (power)", ylab="Scale Free Topology Model Fit,signed R^2", main = "Scale independence")
    text(1:10, 1:10, labels=1:10, col="red")
    abline(h=0.90, col="red")
  })

  output$dendroPlot <- renderPlot({
    # Placeholder for the Dendrogram generated in Step 3
    # plotDendroAndColors(geneTree, dynamicColors, "Dynamic Tree Cut", dendroLabels = FALSE, hang = 0.03, addGuide = TRUE, guideHang = 0.05)
    plot(1:10, main = "Gene Dendrogram and Module Colors")
  })

  output$heatmapPlot <- renderPlot({
    # Placeholder for the Heatmap generated in Step 4
    # labeledHeatmap(Matrix = moduleTraitCor, xLabels = names(clinical_traits), yLabels = names(MEs), ySymbols = names(MEs), colorLabels = FALSE, colors = blueWhiteRed(50), textMatrix = textMatrix, setStdMargins = FALSE, cex.text = 0.5, zlim = c(-1,1), main = paste("Module-trait relationships"))
    image(1:10, 1:10, matrix(1:100, nrow=10), main = "Module-Trait Correlation Heatmap")
  })
}

Summary

By building an R Shiny app, we transformed static analysis results into a dynamic tool for discovery. This allows domain experts to interact with the data directly, fostering better collaboration between bioinformaticians and clinicians.

19.4 Deployment and Packaging Options

Options for making interactive results accessible:

• Shiny Server / shinyapps.io: Host Shiny apps for live, server-rendered interactivity (requires active R session).
• Static HTML widgets: Use htmlwidgets (e.g., plotly, DT) embedded in R Markdown HTML reports for zero-dependency sharing.
• Pre-computed RData: Package pre-computed .RData or .rds files so the app loads instantly without re-running analyses.

Choose the approach that fits your audience; for broad dissemination, static HTML reports with embedded widgets are often easiest to share.