如何在R语言中结合ggballoonplot添加树状图并重构气泡图

Great question! While ggballoonplot doesn’t have built-in dendrogram support like corrplot, you can absolutely add one and reorder your plot based on clustering by combining basic clustering steps with ggplot2’s flexibility. Here’s a step-by-step solution tailored to your use case:

Step 1: Prepare Data and Run Clustering #

First, we’ll cluster either your marker (y-axis) or cluster (x-axis) groups using the expression data, then extract the sorted order from the dendrogram. Let’s use marker groups as an example:

# Load required packages
library(tidyverse)
library(ggpubr)  # For ggballoonplot
library(ggdendro)
library(gridExtra)

# Simulate a dataset matching your structure (skip if you have your own data)
set.seed(123)
X <- expand.grid(cluster = paste0("Cluster_", 1:4), marker = paste0("Marker_", 1:6))
X$Relative.Expression <- runif(nrow(X), 0, 1)
X$Median.Intensity <- sample(c("Low", "Medium", "High"), nrow(X), replace = TRUE)

# Convert data to wide format for clustering (markers as rows, clusters as columns)
wide_data <- X %>% 
  pivot_wider(names_from = cluster, values_from = Relative.Expression) %>% 
  column_to_rownames("marker")

# Run hierarchical clustering
dist_matrix <- dist(wide_data, method = "euclidean")
hc <- hclust(dist_matrix, method = "ward.D2")

# Extract sorted marker order from the dendrogram
sorted_markers <- rownames(wide_data)[hc$order]

# Convert marker to a factor with the sorted order (this reorders the y-axis)
X$marker <- factor(X$marker, levels = sorted_markers)

Step 2: Plot the Reordered Balloon Plot #

Now use ggballoonplot with the sorted factor levels to get your reordered bubble chart:

p_balloon <- ggballoonplot(X, 
                           x = "cluster", 
                           y = "marker", 
                           size = "Relative.Expression", 
                           fill = "Median.Intensity", 
                           ggtheme = theme_classic()) +
  theme(axis.text.y = element_text(size = 10),
        plot.margin = margin(5, 10, 5, 5)) # Adjust margin to align with dendrogram

Step 3: Generate and Combine the Dendrogram #

Use ggdendro to create a ggplot-compatible dendrogram, then combine it with your balloon plot using grid.arrange:

# Create rotated dendrogram (matches the y-axis of the balloon plot)
dendro_plot <- ggdendrogram(hc, rotate = TRUE, theme_dendro = FALSE) +
  theme_void() +
  theme(plot.margin = margin(5, 0, 5, 10)) # Align margin with the balloon plot

# Combine the two plots side-by-side
grid.arrange(dendro_plot, p_balloon, ncol = 2, widths = c(1, 4))

For Clustering Clusters (X-axis) #

If you want to reorder the cluster groups instead, follow the same logic but adjust the clustering and dendrogram orientation:

# Cluster clusters instead of markers
wide_data_clusters <- X %>% 
  pivot_wider(names_from = marker, values_from = Relative.Expression) %>% 
  column_to_rownames("cluster")

dist_clusters <- dist(wide_data_clusters, method = "euclidean")
hc_clusters <- hclust(dist_clusters, method = "ward.D2")
sorted_clusters <- rownames(wide_data_clusters)[hc_clusters$order]

# Update cluster factor order
X$cluster <- factor(X$cluster, levels = sorted_clusters)

# Balloon plot with sorted clusters
p_balloon_x <- ggballoonplot(X, 
                           x = "cluster", 
                           y = "marker", 
                           size = "Relative.Expression", 
                           fill = "Median.Intensity", 
                           ggtheme = theme_classic()) +
  theme(axis.text.x = element_text(angle = 45, hjust = 1),
        plot.margin = margin(10, 5, 5, 5))

# Top-aligned dendrogram
dendro_plot_x <- ggdendrogram(hc_clusters, rotate = FALSE, theme_dendro = FALSE) +
  theme_void() +
  theme(plot.margin = margin(5, 5, 0, 5))

# Combine plots vertically
grid.arrange(dendro_plot_x, p_balloon_x, nrow = 2, heights = c(1, 4))

The core idea here is to first use clustering to define the order of your groups, convert those groups to factors with the sorted order, then build the dendrogram as a separate ggplot object and combine it with the balloon plot. This gives you the same clustered ordering effect as corrplot, but with the full flexibility of ggplot2.

内容的提问来源于stack exchange，提问作者startswithH