R语言树状图：截取顶部11分支并实现叶节点均匀分布

问题背景 #

需要处理含1000余个叶节点的树状图，目标是保留顶部11个分支，同时让新生成的叶节点在树状图末端水平均匀分布。之前尝试通过裁剪x轴范围实现，但裁剪后叶节点间距仍基于原始20个节点的布局，无法达到均匀分布的效果。测试代码如下：

# Set seed for reproducibility
set.seed(123)

# Create a data frame with 20 rows and 25 columns of random data
test_data <- as.data.frame(matrix(rnorm(20 * 25), nrow = 20, ncol = 25))

# Rename columns to stat1, stat2, ..., stat25
colnames(test_data) <- paste0("stat", 1:25)

# Scale the data (mean = 0, SD = 1)
scaled_data <- scale(test_data)

# Calculate the distance matrix
dist_matrix <- dist(scaled_data, method = "euclidean")

# Perform hierarchical clustering
hc <- hclust(dist_matrix, method = "ward.D2")

# Convert the `hclust` object into a `phylo` object for ggtree
phylo_tree <- as.phylo(hc)

# Cut the hclust tree into 11 clusters
clusters <- cutree(hc, k = 11)

# Add cluster information to a data frame that matches the tips of the tree
tip_data <- data.frame(
  label = phylo_tree$tip.label,  # Tree tip labels
  cluster = clusters  # Cluster assignments
)

# Plot with tip coloring by cluster
ggtree_plot <- ggtree(phylo_tree, layout = "rectangular", branch.length = "height") %<+% tip_data +
  geom_tiplab(aes(label = cluster), size = 2.5, align = TRUE, offset = 0.05) +
  geom_tippoint(aes(color = factor(cluster)), size = 3) +
  scale_color_manual(values = rainbow(11)) +  # Use a palette with 11 distinct colors
  theme_tree2()

print(ggtree_plot)

### CROP the tree to a height with 11 branches

# Modify the ggtree plot to limit the x-axis range
ggtree_plot <- ggtree(phylo_tree, layout = "rectangular", branch.length = "height") %<+% tip_data +
  scale_color_manual(values = rainbow(11)) +  # Use a palette with 11 distinct colors
  theme_tree2() +
  coord_cartesian(xlim = c(0, 2.0))  # Set the x-axis range to 0-2.0 in this case with seed 123

print(ggtree_plot)

核心问题分析 #

直接裁剪x轴只是隐藏了部分分支，但树的底层结构仍基于原始所有叶节点，因此节点间距不会自动调整。要实现均匀分布，必须重新构建仅包含11个分支的树结构，将每个簇合并为单个叶节点。

解决方案代码 #

set.seed(123)
library(ape)
library(ggtree)
library(phytools)

# 生成测试数据（替换为你的1000+节点数据集即可）
test_data <- as.data.frame(matrix(rnorm(20 * 25), nrow = 20, ncol = 25))
colnames(test_data) <- paste0("stat", 1:25)
scaled_data <- scale(test_data)
dist_matrix <- dist(scaled_data, method = "euclidean")
hc <- hclust(dist_matrix, method = "ward.D2")
phylo_tree <- as.phylo(hc)

# 切割为11个簇
k <- 11
clusters <- cutree(hc, k = k)

# 将簇信息转为命名向量，用于合并节点
cluster_vec <- clusters
names(cluster_vec) <- phylo_tree$tip.label

# 合并同一簇的所有叶节点，生成仅含11个节点的新树
merged_tree <- mergeTips(phylo_tree, cluster_vec, type = "clade")

# 绘制新树，叶节点自动均匀分布
ggtree(merged_tree, layout = "rectangular", branch.length = "height") +
  geom_tiplab(aes(label = label), size = 3) +
  geom_tippoint(aes(color = factor(label)), size = 3) +
  scale_color_manual(values = rainbow(k)) +
  theme_tree2()

方案说明 #

1. 合并节点：使用phytools包的mergeTips函数，将原树中属于同一簇的所有叶节点合并为单个节点，生成仅包含11个叶节点的新树结构。
2. 均匀分布：新树的布局完全基于11个节点，绘制时叶节点会自动水平均匀分布，不再受原始1000+节点的间距影响。
3. 扩展性：该方法适用于任何规模的原始数据集，只需替换测试数据部分为你的实际数据即可。

内容的提问来源于stack exchange，提问作者R student