vignettes/Introduction.R

## ------------------------------------------------------------------------
suppressMessages(library(dendextend))
library(d3heatmap)

## ------------------------------------------------------------------------
x <- mtcars # [c(2:4,7),1:4]

d3heatmap(x, k_row = 4, k_col = 2)
d3heatmap(x, k_row = 4, k_col = 2, scale = "column") # a more appropriate scalling

# row_dend2 <- x %>% dist %>% hclust %>% as.dendrogram %>%
#   color_branches(k = 3)
# col_dend2 <- x %>% t %>% dist %>% hclust %>% as.dendrogram %>%
#   color_branches(k = 2)
# d3heatmap(x, Rowv = row_dend2, Colv = col_dend2)
# d3heatmap(x, Rowv = row_dend2, Colv = col_dend2, scale = "column") # a more appropriate scaling

## ------------------------------------------------------------------------
x_cor <- cor(x)

x_cor %>% d3heatmap(k_row = 3, k_col = 3) # (symm = TRUE)
# x_cor %>% d3heatmap(reorderfun = function(x, ...) x) # this doesn't reorder the rows/columns based on the means.
# Note the use of k_row/k_col to color the branches

# # reproducing the dendrogram, but with manually created the dendrograms
# cor_dend <- x_cor  %>% dist %>% hclust(method = "com") %>% as.dendrogram %>%
#   color_branches(k = 3) %>% 
#   reorder(rowMeans(x_cor)) # this is what is done inside heatmap.
# x_cor %>% d3heatmap(Rowv = cor_dend, Colv = cor_dend)



## ------------------------------------------------------------------------
x <- datasets::iris[,-5] # [c(2:4,7),1:4]

row_dend2 <- x %>% dist %>% hclust(method = "ave") %>% as.dendrogram %>%
  color_branches(k = 3) %>% 
  reorder(rowMeans(x)) 
col_dend2 <- x %>% t %>% dist %>% hclust %>% as.dendrogram %>%
  color_branches(k = 2) %>% 
  sort # This sorting requires dendextend and cannot be easily done without it

d3heatmap(x, Rowv = row_dend2, Colv = col_dend2,
          colors = "Greens",
          # scale = "row",
          width = 600,
          show_grid = FALSE)

## ---- cache=FALSE--------------------------------------------------------
sessionInfo()
rstudio/d3heatmap documentation built on Nov. 18, 2024, 9:20 a.m.