Clusters can be found in a data set by chance due to clustering noise or sampling error. This article describes the R package pvclust<\/strong> (Suzuki and Shimodaira 2015)<\/span> which uses bootstrap resampling techniques to compute p-value<\/strong> for each hierarchical clusters<\/strong>.<\/p>\n Contents:<\/p>\n Clusters with AU >= 95% are considered to be strongly supported by data.<\/p>\n<\/div>\n<\/div>\n We\u2019ll use lung<\/em> data set [in pvclust<\/em> package]. It contains the gene expression profile of 916 genes of 73 lung tissues including 67 tumors. Columns are samples and rows are genes.<\/p>\n We\u2019ll use only a subset of the data set for the clustering analysis. The R function sample<\/em>() can be used to extract a random subset of 30 samples:<\/p>\n The function pvclust<\/em>() can be used as follow:<\/p>\n Note that, the computation time can be strongly decreased using parallel computation version called parPvclust<\/em>(). (Read ?parPvclust() for more information.)<\/p>\n The function pvclust<\/em>() returns an object of class pvclust<\/em> containing many elements including hclust<\/em> which contains hierarchical clustering result for the original data generated by the function hclust<\/em>().<\/p>\n<\/div>\n pvclust<\/em>() performs clustering on the columns of the data set, which correspond to samples in our case. If you want to perform the clustering on the variables (here, genes) you have to transpose the data set using the function t<\/em>().<\/p>\n The R code below computes pvclust<\/em>() using 10 as the number of bootstrap replications (for speed):<\/p>\n\n
\n
Related Book <\/a><\/h4>Practical Guide to Cluster Analysis in R<\/div>\n
Algorithm<\/h2>\n
\n
\n
Required packages<\/h2>\n
\n
install.packages(\"pvclust\")<\/code><\/pre>\n\n
library(pvclust)<\/code><\/pre>\n<\/div>\nData preparation<\/h2>\n
library(pvclust)\r\n# Load the data\r\ndata(\"lung\")\r\nhead(lung[, 1:4])<\/code><\/pre>\n## fetal_lung 232-97_SCC 232-97_node 68-96_Adeno\r\n## IMAGE:196992 -0.40 4.28 3.68 -1.35\r\n## IMAGE:587847 -2.22 5.21 4.75 -0.91\r\n## IMAGE:1049185 -1.35 -0.84 -2.88 3.35\r\n## IMAGE:135221 0.68 0.56 -0.45 -0.20\r\n## IMAGE:298560 NA 4.14 3.58 -0.40\r\n## IMAGE:119882 -3.23 -2.84 -2.72 -0.83<\/code><\/pre>\n# Dimension of the data\r\ndim(lung)<\/code><\/pre>\n## [1] 916 73<\/code><\/pre>\nset.seed(123)\r\nss <- sample(1:73, 30) # extract 20 samples out of\r\ndf <- lung[, ss]<\/code><\/pre>\n<\/div>\nCompute p-value for hierarchical clustering<\/h2>\n
Description of pvclust() function<\/h3>\n
pvclust(data, method.hclust = \"average\",\r\n method.dist = \"correlation\", nboot = 1000)<\/code><\/pre>\nparPvclust(cl=NULL, data, method.hclust = \"average\",\r\n method.dist = \"correlation\", nboot = 1000,\r\n iseed = NULL)<\/code><\/pre>\n\n
Usage of pvclust() function<\/h3>\n
library(pvclust)\r\nset.seed(123)\r\nres.pv <- pvclust(df, method.dist=\"cor\", \r\n method.hclust=\"average\", nboot = 10)<\/code><\/pre>\n# Default plot\r\nplot(res.pv, hang = -1, cex = 0.5)\r\npvrect(res.pv)<\/code><\/pre>\n![](\"https:\/\/www.datanovia.com\/en\/wp-content\/uploads\/dn-tutorials\/004-cluster-validation\/figures\/018-p-value-for-hierarchical-clustering-pvclust-p-value-hierarchical-clustering-1.png\")
<\/p>\n