Question

Upper triangular matrix heatmap using pheatmap package

0

Entering edit mode

Deepak Tanwar ▴ 10

@deepak-tanwar-8113

Last seen 4 months ago

McGill University, Canada

How can I plot only upper triangular portion of a gene by gene correlation matrix with pheatmap package ?

Even if I try to remove lower triangle by: mat[lower.tri(mat)] <- "NA" it does not work.

pheatmap • 8.5k views

ADD COMMENT • link 8.9 years ago Deepak Tanwar ▴ 10

1

Entering edit mode

It probably will work if you remove quotation marks around NA. Also turn off the clustering of rows and columns to see nice triangular plot.

ADD REPLY • link 8.9 years ago Raivo Kolde ▴ 50

0

Entering edit mode

It does work, but I want the rows and columns to be clustered.

ADD REPLY • link 8.9 years ago Deepak Tanwar ▴ 10

1

Entering edit mode

Then you have to cluster the data before you assign some values NA. For example assuming that mat is full correlation matrix you can do the following

hc = hclust(as.dist(1 - mat))
mat = mat[hc$order, hc$order]
mat[lower.tri(mat)] = NA

pheatmap(mat, cluster_col = F, cluster_row = F)

ADD REPLY • link 8.9 years ago Raivo Kolde ▴ 50

0

Entering edit mode

It's perfect Raivo.

Is there a way to also show dendogram along with heatmap?

I tried

pheatmap(mat, cluster_col = hc$order, cluster_row = hc$order)

ADD REPLY • link 8.9 years ago Deepak Tanwar ▴ 10

1

Entering edit mode

It is a hack but basically you can achieve it like this (starting from full correaltion matrix again)

o = rownames(mat)
hc = hclust(as.dist(1 - mat))
mat = mat[hc$order, hc$order]
mat[lower.tri(mat)] = NA
mat = mat[o, o]

pheatmap(mat, cluster_col = hc, cluster_row = hc)

ADD REPLY • link 8.9 years ago Raivo Kolde ▴ 50

0

Entering edit mode

I tried to perform the code, but I am getting following error. I also tried to look for the error:

Error in if (cluster_rows) { : argument is not interpretable as logical
In addition: Warning message:
In if (cluster_rows) { :
  the condition has length > 1 and only the first element will be used

ADD REPLY • link 8.9 years ago Deepak Tanwar ▴ 10

0

Entering edit mode

Hi,

I am not yet able to figure out that:

test = matrix(rnorm(200), 20, 10)

test[1:10, seq(1, 10, 2)] = test[1:10, seq(1, 10, 2)] + 3

test[11:20, seq(2, 10, 2)] = test[11:20, seq(2, 10, 2)] + 2

test[15:20, seq(2, 10, 2)] = test[15:20, seq(2, 10, 2)] + 4

colnames(test) = paste("Test", 1:10, sep = "")

rownames(test) = paste("Gene", 1:20, sep = "")



mat <- cor(t(test))
o = rownames(mat)
hc = hclust(as.dist(1 - mat))
mat = mat[hc$order, hc$order]
mat[lower.tri(mat)] = NA
mat = mat[o, o]

pheatmap(mat, cluster_rows = hc, cluster_cols = hc)

Error in if (cluster_rows) { : argument is not interpretable as logical In addition: Warning message: In if (cluster_rows) { : the condition has length > 1 and only the first element will be used

ADD REPLY • link 8.9 years ago Deepak Tanwar ▴ 10

1

Entering edit mode

You have to update your version of pheatmap. This feature of giving clustering as a parameter was introduces fairly recently.