如何在R中对数据矩阵进行分层聚类?

问题描述

我正在尝试对由科学数据产生的数据矩阵进行聚类.我知道我要如何完成聚类，但是不确定如何在R中完成这一壮举.

I am trying to cluster a data matrix produced from scientific data. I know how I want the clustering done, but am not sure how to accomplish this feat in R.

这是数据的样子:

            A1     A2     A3     B1     B2     B3     C1     C2     C3
sample1      1      9     10      2      1     29      2      5     44
sample2      8      1     82      2      8      2      8      2     28
sample3      9      9     19      2      8      1      7      2     27

请考虑A1，A2，A3是一次处理的三个重复，以及B和C.Sample1是不同的测试变量.因此，我想对该矩阵进行分层聚类，以查看列之间的所有差异，特别是我将制作树状图(树)以观察列的相关性.

Please consider A1,A2,A3 to be three replicates of a single treatment, and likewise with B and C. Sample1 are different tested variables. So, I want to hierarchically cluster this matrix in order to see the over all differences between the columns, specifically I will be making a dendrogram (tree) to observe the relatedness of the columns.

有人知道如何适当地聚类这样的东西吗?我尝试这样做:

Does anyone know how to appropriately cluster something like this? I tried doing this with this:

raw.data <- read.delim("test.txt",header=FALSE,stringsAsFactors=FALSE)
dist.mat<-vegdist(raw.data,method="jaccard")
clust.res<-hclust(dist.mat)
plot(clust.res)

...但是，这导致了每个样本变量而不是每个列都有分支的树.感谢您的任何建议！

...However, this resulted in a tree with branches for each sample variable, instead of each column. Thanks for any suggestions!

推荐答案

只需转置您的数据集:

raw.data <- t(raw.data)
require(vegan)
dist.mat<-vegdist(raw.data,method="jaccard")
clust.res<-hclust(dist.mat)
plot(clust.res)

这篇关于如何在R中对数据矩阵进行分层聚类?的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持！