如何在ggplot2带层次聚类树的百分比堆叠柱状图中添加分类变量

1. 准备示例数据&加载依赖包 #

先定义符合你场景的简化数据集，同时加载需要的R包：

library(ggplot2)
library(ggdendro)
library(dplyr)
library(tidyr)
library(patchwork)

# 示例数据集：患者1、3为HIV+，患者2为HIV-
sample_data <- data.frame(
  Sample = c("患者1", "患者2", "患者3"),
  HIV_Status = factor(c("HIV+", "HIV-", "HIV+")),
  CellType_A = c(30, 50, 25),
  CellType_B = c(40, 30, 45),
  CellType_C = c(30, 20, 30)
)

# 转换为堆叠柱状图需要的长格式
long_data <- pivot_longer(sample_data, 
                          cols = starts_with("CellType"), 
                          names_to = "CellType", 
                          values_to = "Percentage")

2. 生成层次聚类树并关联样本信息 #

计算聚类后，提取树的绘图数据，同时把HIV状态关联到树的节点上：

# 提取聚类用的数值矩阵
cluster_matrix <- sample_data %>% 
  select(starts_with("CellType")) %>% 
  as.matrix()
rownames(cluster_matrix) <- sample_data$Sample

# 计算层次聚类
hc <- hclust(dist(cluster_matrix))

# 提取树的绘图数据，关联HIV状态
dendro_data <- dendro_data(hc, type = "rectangle")
node_info <- dendro_data$labels %>% 
  left_join(sample_data %>% select(Sample, HIV_Status), 
            by = c("label" = "Sample"))

3. 绘制带样本名称的聚类树 #

添加样本名称到树的节点旁（优先级低的需求），调整参数避免标签重叠：

p_dendro <- ggplot() +
  # 绘制树的分支
  geom_segment(data = dendro_data$segments, 
               aes(x = x, y = y, xend = xend, yend = yend)) +
  # 添加样本名称
  geom_text(data = dendro_data$labels, 
            aes(x = x, y = y, label = label), 
            hjust = -0.1, size = 3) +
  theme_dendro() +
  coord_flip() +
  ylim(-5, max(dendro_data$segments$y) + 10) # 预留标签空间

4. 绘制关联HIV状态的堆叠柱状图 #

这里提供两种方式关联HIV分类颜色：

方式1：柱子边框对应HIV状态颜色

# 按聚类顺序排序样本
sample_order <- dendro_data$labels$label
long_data$Sample <- factor(long_data$Sample, levels = sample_order)

p_bar <- ggplot(long_data %>% left_join(sample_data), 
                aes(x = Sample, y = Percentage, fill = CellType)) +
  geom_col(position = "fill", aes(color = HIV_Status), linewidth = 1.2) +
  scale_y_continuous(labels = scales::percent) +
  scale_color_manual(values = c("HIV+" = "#e74c3c", "HIV-" = "#3498db")) +
  scale_fill_brewer(palette = "Set2") +
  coord_flip() +
  theme_minimal() +
  theme(legend.position = "bottom")

方式2：柱子顶部添加HIV状态颜色标记

p_bar <- ggplot(long_data, aes(x = Sample, y = Percentage, fill = CellType)) +
  geom_col(position = "fill", color = "black") +
  # 在柱子顶部添加对应HIV状态的颜色点
  geom_point(data = sample_data, 
             aes(x = Sample, y = 1.1, color = HIV_Status), 
             size = 5) +
  scale_y_continuous(labels = scales::percent, limits = c(0, 1.2)) +
  scale_color_manual(values = c("HIV+" = "#e74c3c", "HIV-" = "#3498db")) +
  scale_fill_brewer(palette = "Set2") +
  coord_flip() +
  theme_minimal() +
  theme(legend.position = "bottom")

5. 拼接聚类树与柱状图 #

用patchwork包将两个图拼接，调整宽度比例适配显示：

p_dendro + p_bar + plot_layout(widths = c(1, 2))

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