Hello,

Is there a way to assign a custom (row) dendrogram and ordering to a Heatmap object generated by ComplexHeatmap Heatmap() function?

After finding out about the row_dend() and row_order() functions, I would have thought something like this would work:

library(ComplexHeatmap)

mat <- matrix(rnorm(1000), ncol=10, nrow=10)
# calculate Spearman correlation
mat.cor <- cor(t(mat), method = "spearman")
# calculate distance based on correlation matrix
mat.dist <- as.dist(1 - abs(mat.cor))
# hierarchical clustering
mat.hc <- hclust(mat.dist, method = "ward.D2")
# dendrogram
mat.dend <- as.dendrogram(mat.hc)

# create heatmap
hm <- Heatmap(mat)

# substitute row dendrogram and order
row_dendro(hm)[[1]] <- mat.dend
row_order(hm)[[1]] <- order.dendrogram(mat.dend)

# plot heatmap with custom row dendrogram and ordering
print(hm)

However, I get these errors:

row_dend(hm)[[1]] <- mat.dend
Error in row_dend(hm)[[1]] <- mat.dend :
  could not find function "row_dend<-"
row_order(hm)[[1]] <- order.dendrogram(mat.dend)
Error in row_order(hm)[[1]] <- order.dendrogram(mat.dend) :
  could not find function "row_order<-"

The two functions work fine when called on the Heatmap object, it's only the assignment that fails.

Thank you!

`mat.dend` can be assigned to `cluster_rows` argument:

Heatmap(mat, cluster_rows = mat_dend)

which means `cluster_rows` can be a logical value which controls whether do clustering, it can also be a `hclust` or a `dendrogram` object which already contains a customized clustering object.