I used deeptools to make a heatmap of a ChIP-seq experiment with 4 samples - WT_TF1, WT_TF2, Mutant_TF1, and Mutant TF2. I made the heatmap referenced to a bedfile of all unique peak sites for all the conditions. I used the --kmeans option of plotHeatmap in an attempt to do a clustering analysis whereby the clusters note enrichment of binding in some samples but not all (For example, cluster 1 could show regions where all samples bind, cluster 2 shows regions where sample 1 and 2 bind but not sample 3, etc). I wanted to know if there was a way of graphing the different clusters similar to the way it was done in Miranda et al (the figure on the right side). Attached below a picture of my heatmap generated from plotHeatmap and a picture of what was done in the Miranda paper. The Miranda paper used Homer to do the ChIP-seq analysis and it wasn't so clear ho they made the graph next to the heatmap. I was wondering if anyone knew of a way I could generate the graph with my data.
Thank you so much,