| The heatmap is produced by using the "hclust" hierarchical clustering function from R. By default, the dissimilarity used is "1-correlation" (with the Pearson correlation coefficient), and the Ward agglomeration method. Other distances (e.g., "euclidean") or agglomeration methods (e.g., "average") can be used. For visulization, a scaling argument is suggested by default to enhance contrast (does not modify the clustering computation). Note that computation times for big matrices (e.g. with thousands of features) can be long. |