Question: How to reduce the length (size) of the dengrogram in heatmap.2
1
Wox340 wrote:

This is the code I've used to build by Heatmap with dendrogram

``````heatmap.2(as.matrix(SI_Ratios), scale = "none", col = bluered(100),
trace = "none", density.info = "none", cexRow=0.5, cexCol=0.75, margins = c(5,20),
key.xlab = "Ln(Peak Area / T0)")
`````` How can I reduce the length of the dendrogram? Also How to increase the font size of the Color Key description?

R gene • 6.8k views
modified 2.5 years ago by Kevin Blighe66k • written 2.5 years ago by Wox340
1
Kevin Blighe66k wrote:

For modifying the font size of the colour key, just use the `key.lab` parameter (as I do below).

For changing the 'length' of the dendrogram, you have to realise that the dendrogram heights are representative of the distances between your samples and are measured on the scale that you used during the creation of the distance matrix, usually Euclidean distance. Euclidean distances can be large, depending on your data-type, even into the 100s.

You can play around with the margins in order to 'squeeze' the dendrgrams - you can do this via `par(oma)`. Here, I have modified the left margin with `par(oma=c(1,15,1,1))`:

``````data <- replicate(20, rnorm(50))
rownames(data) <- paste("Gene", c(1:nrow(data)))
colnames(data) <- paste("Sample", c(1:ncol(data)))
par(oma=c(1,15,1,1)); heatmap.2(data, key=TRUE, keysize=1.5, key.xlab="X-axis", key.ylab="", key.title="", key.par=list(mar=c(4,0,1,2), cex=1.0, cex.lab=2.0, cex.axis=2.0), margins=c(5,5))
`````` If you wanted a distance metric that is measured on a smaller range, you could do correlation distance by adding `distfun=function(x) as.dist(1-cor(t(x)))` to the `heatmap.2` function, as I do here: A: Hierarchical Clustering in single-channel agilent microarray experiment

Kevin

Thanks, Kevin,

This works for me. However, I can not zoom the Heatmap and its blurred. How can I save this to a PNG with good resolution

2

Don't use PNG if you want a very clear image. PNG, TIFF, BMP, etc., are pixel-based. If you instead use a vector-based graphics format, then you would get publication-ready figures. You can then use the width and height parameters in order to modify the size (inches) of the output.

``````#Default size
pdf("Heatmap.pdf", width=7, height=7)
...
dev,.off()

#Wide heatmap
pdf("Heatmap.pdf", width=9, height=4)
...
dev,.off()

#Skinny heatmap
pdf("Heatmap.pdf", width=4, height=9)
...
dev,.off()
``````