基于标签重新排序和着色树状图

Question

我正在尝试控制树状图的顺序和颜色。显然树状图的要点是按相似性排序，但在分支内我想设置一个有意义的顺序（字母数字）。

library(vegan)
library(stats) 

x <-data.frame(data = c(1:10)) 
y = data.frame(type = c("A","B","C","A","C","D","A","B","C","B"), site_name = c("A1","B1","C1","A2","C2","D1","A3","B2","C3","B3"))  
row.names(x) = y$site_name 
dis = vegdist(x) 
hc <- hclust(dis) 
dd <- as.dendrogram(hc) 
plot(dd)

我的数据标签是文本，但它们确实在变量中列出了固定顺序

site_order = c("A1","A2","A3","B1","B2","B3","C1","C2","C3","D1")

1) 我想找到一个解决方案，根据分支内的 site_order 对树状图进行排序。

e.g. A1,B1, A2, C1, C2,D2, A3,B2, B3,C3

我还想使用 site_type 为标签着色和塑造形状例如（A= 红色圆圈，B= 蓝色正方形，C= 绿色三角形，D = 黄色十字）

这可能吗？

Answer 1

这项工作最好使用 dendextend 的 rotate 函数来完成。

library(vegan)
library(stats) 

x <-data.frame(data = c(1:10)) 
y = data.frame(type = c("A","B","C","A","C","D","A","B","C","B"), site_name = c("A1","B1","C1","A2","C2","D1","A3","B2","C3","B3"))  
row.names(x) = y$site_name 
dis = vegdist(x) 
hc <- hclust(dis) 
dd <- as.dendrogram(hc) 

par(mfrow = c(1,2))
plot(dd, main = "orig")

library(dendextend)
dd2 <- rotate(dd, sort(labels(dd)) )
plot(dd2, main = "as sorted as possible \n(under the constraints)")

输出：

您可以从 the online vignettes 了解更多关于 dendextend 包的信息。

基于标签重新排序和着色树状图

Reorder and colour dendogram based on labels

r

dendrogram

hclust

dendextend