Unsupervised Hierarchical Cluster Heatmap

Question

[esantosh10]

Hello All,

Please look at the data:

Sample  Sample1     Sample2     Sample3     Sample4     Sample5

ABC1    0.68635286  0.48764939  0.33512254  0.01522009  0.7361551

ABC2    0.05934282  0.4897348   0.09595364  0.1200324   0.88306627

ABC3    0.42620642  0.5320958   0.74631177  0.63651979  0.31220708

I have multiple samples vs multiple genes categorized into two categories of normal and tumor samples. I would like to generate a heat map as the below image i.e Genes vs Samples(Normal vs Tumors):

I have around 1000 genes for both the Normal vs Tumor Datasets and the number of samples are 120 for tumor and 100 for Normal using ggplot or R? I tried using the hclust method but the plot looks very bad. Any other ways to plot such high number of rows and columns ?

[GenoMax]

Peter Cock's heatmap tutorial.

[Jean-Karim Heriche]

Also this post.

[5heikki]

That's a good post. It's missing the top clustering algorithm though, i.e. Affinity Propagation Clustering

[Kevin Blighe]

To add another comment:

I tried using the hclust method but the plot looks very bad.

What is your definition of "bad"? - something that's not symmetrical?; something that doesn't fit your hypothesis?