top_supervised: A Function for Assisting Supervised Clustering

In AutoPipe: Automated Transcriptome Classifier Pipeline: Comprehensive Transcriptome Analysis

Description

when perfoming a supervised clustering the user should run this function in order to get the best results.

Usage

top_supervised(me,TOP=1000,cluster_which,TRw=-1)

Arguments

me	the matrix of the gene exporessions, the olums should be the samples and the colnames the sample names the rownames should be the genes . at best the ENTEREZID
TOP	the top genes to choose, default is 100.
cluster_which	a dataframe with the supervised clustering arrangment of the samples. the dataframe should have the sample names in the first column and the clustering in the secound column.
TRw	the threshhold for excluding samples with silhouette width < TRw

Value

a list. the first place is the expression matrix, the secound is the silhouette for each sample.

Examples

library(org.Hs.eg.db)
data(rna)
cluster_which<-cbind(colnames(rna),c(rep(1,times=24),rep(2,times=24)))
me_x=rna
## calculate best number of clusters and
res<-top_supervised(me_x,TOP = 100,cluster_which)
me_TOP=res[[1]]
number_of_k=2
groups_men=res[[2]]
me_x=me_TOP
colnames(me_x)
o_g<-Supervised_Cluster_Heatmap(groups_men = groups_men, gene_matrix=me_x,
                               method="PAMR",show_sil=TRUE,print_genes=TRUE,threshold = 0,
                               TOP = 100,GSE=FALSE,plot_mean_sil=FALSE,stats_clust=res[[2]],
                               samples_data = as.data.frame(groups_men[,1,drop=FALSE]))
                               

AutoPipe documentation built on May 1, 2019, 7:28 p.m.