如何在根节点合并多个hclust对象或树状图？附USArrests聚类示例

当然有办法啦！要把多个hclust对象在根节点合并，核心思路是先把hclust转换成系统发育树格式（比如phylo对象），然后合并这些树，最后再转回hclust或者直接用于热图展示。下面结合你的USArrests数据集示例一步步来：

1. 准备数据与地区划分 #

先补全完整的美国地区划分，给数据集添加地区标签：

# 加载必要工具包
library(ape)
library(pheatmap)
library(dplyr)

# 完整的美国四大州地区划分
Northeast <- c("Connecticut", "Maine", "Massachusetts", "New Hampshire", "Rhode Island", "Vermont", "New Jersey", "New York", "Pennsylvania")
Midwest <- c("Illinois", "Indiana", "Michigan", "Ohio", "Wisconsin", "Iowa", "Kansas", "Minnesota", "Missouri", "Nebraska", "North Dakota", "South Dakota")
South <- c("Delaware", "Florida", "Georgia", "Maryland", "North Carolina", "South Carolina", "Virginia", "West Virginia", "Alabama", "Kentucky", "Mississippi", "Tennessee", "Arkansas", "Louisiana", "Oklahoma", "Texas")
West <- c("Arizona", "Colorado", "Idaho", "Montana", "Nevada", "New Mexico", "Utah", "Wyoming", "Alaska", "California", "Hawaii", "Oregon", "Washington")

# 给USArrests添加地区列
USArrests_regions <- USArrests %>%
  mutate(Region = case_when(
    rownames(.) %in% Northeast ~ "Northeast",
    rownames(.) %in% Midwest ~ "Midwest",
    rownames(.) %in% South ~ "South",
    rownames(.) %in% West ~ "West"
  ))

2. 按地区分别做层次聚类 #

对每个地区的子集单独计算聚类，得到各自的hclust对象：

# 按地区拆分数据集并逐个聚类
cluster_list <- USArrests_regions %>%
  group_split(Region) %>%
  lapply(function(df) {
    # 提取数值型数据并标准化
    scaled_data <- scale(df %>% select(-Region))
    # 计算距离矩阵并执行层次聚类
    dist_mat <- dist(scaled_data)
    hclust(dist_mat, method = "ward.D2")
  })

# 给聚类列表命名，方便后续识别
names(cluster_list) <- c("Northeast", "Midwest", "South", "West")

3. 在根节点合并多个hclust对象 #

原生hclust结构不支持直接合并，我们先转成phylo树格式，再用bind.tree在根节点合并：

# 将每个hclust对象转换为phylo系统发育树
phylo_list <- lapply(cluster_list, as.phylo)

# 初始化合并树为第一个地区的树，然后依次合并剩余树到根节点
merged_tree <- phylo_list[[1]]
for (i in 2:length(phylo_list)) {
  # position=0 表示在根节点处合并
  merged_tree <- bind.tree(merged_tree, phylo_list[[i]], position = 0)
}

# 如果需要转回hclust格式（比如给pheatmap用），直接用as.hclust转换
merged_hclust <- as.hclust(merged_tree)

4. 用合并后的聚类树绘制热图 #

现在可以把合并后的树作为行聚类规则，搭配地区注释绘制热图：

# 准备行注释数据（标注每个州所属地区）
annotation_row <- USArrests_regions %>% select(Region)
rownames(annotation_row) <- rownames(USArrests_regions)

# 绘制热图，指定行聚类为合并后的hclust对象
pheatmap(scale(USArrests), 
         cluster_rows = merged_hclust,  # 使用合并后的聚类树
         annotation_row = annotation_row,
         main = "USArrests Clustered by Region (Merged at Root)",
         treeheight_row = 60)

关键细节说明 #

• 为什么要转phylo格式？因为原生hclust的结构只记录单棵树的聚类路径，不支持多树合并操作；而phylo是专门用于树结构的格式，支持灵活的树拼接。
• bind.tree的position=0是核心参数，确保每个地区的聚类树都作为合并后大树的一级分支，完美实现根节点合并的需求。

内容的提问来源于stack exchange，提问作者csgroen