R Notebook

raw<-read.table("G:\\My Drive\\document for each subject\\senior_project\\67\\ploy_sp.csv", header=TRUE, sep = ",")

library(vegan)

sp<-raw[1:18,2:ncol(raw)]

sp_log <- log1p (sp)  # log-transform raw species composition data
dist_sp <- vegdist (sp_log, method = 'bray') # calculate Bray-Curtis distance matrix

Warning: you have empty rows: their dissimilarities may be
                 meaningless in method “bray”

dist_sp_sqrt <- sqrt (dist_sp) # square-root BC distance matrix to make distances metric
clus_ward <- hclust (dist_sp_sqrt, method = 'ward.D2')  # calculate Ward's algorithm (using the correct ''method = 'ward.D2''')
clus_ward_cut <- cutree (clus_ward, k = 3) # "cut the tree" - to which groups individual samples belong?

plot (clus_ward, cex = .5)  # argument cex reduced the size of the dendrogram leaf labels to make them readable
clus_in_dendro <- unique (clus_ward_cut[clus_ward$order]) # make sure to know which box is which cluster!
rect.hclust (clus_ward, k = 3, border = clus_in_dendro)

legend ('topleft', legend = paste ('Cluster', 1:3), col = 1:3, pch = 22, bty = 'n')