Pheatmap: Cluster columns on hclust object

Question

[michaeljon.rosslee]

Hi

Problem with cluster_col in pheatmap.

I am attempting to pass an hclust object (hc) to pheatmap and have found that it will correctly construct the dendrogram on the clusters, but the order of the tiles/columns in the actual heatmap is incorrect.

pheatmap code:

 heat <- pheatmap(
    mat               = t(data),
    color             = viridis(10),
    breaks            = group_breaks,
    annotation_col    = group_col,
    annotation_colors = group_col_colours,
    cluster_cols      = hc,
    cluster_rows      = FALSE, 
    border_color      = NA,
    show_colnames     = TRUE,
    drop_levels       = TRUE,
    main              = "Heatmap"
)

Dendrogram:

Column order:

When I investigate hc$tree and heat$tree_col$order this matches the order of the columns:

hc$order
 [1] 31  3  9 28  1 23 12 16  4 29 11  2  5 27  6 30 19 20 14 17 10 21 18 22 13 25  8 26 24
[30]  7 15 54 35 50 38 47 34 37 40 43 51 53 49 52 39 44 42 48 41 45 33 36 32 46

heat$tree_col$order
 [1] 31  3  9 28  1 23 12 16  4 29 11  2  5 27  6 30 19 20 14 17 10 21 18 22 13 25  8 26 24
[30]  7 15 54 35 50 38 47 34 37 40 43 51 53 49 52 39 44 42 48 41 45 33 36 32 46

Which makes sense given that in the function tree_col is set to hc$order.

In short, how can I ensure that the column order of pheatmap matches the displayed order of the dendrogram (as shown in above images?) in pheatmap? I have tried ordering the dendrogram (dendsort), but to no avail.

[1769mkc]

this might of some help extract dendrogram cluster from pheatmap

Answer 1

[Kevin Blighe]

I am not sure what you mean. When I do it, they align:

#Create random data
data <- replicate(20, rnorm(50))
rownames(data) <- paste("Gene", c(1:nrow(data)))
colnames(data) <- paste("Sample", c(1:ncol(data)))

hc <- hclust(dist(t(data)))
plot(hc)

out <- pheatmap(data, cluster_cols = hc)

out$tree_col$order
 [1] 18  2  9 16 17 19 15  1 14 20  3  7 11 13  5 12  4  6  8 10