Process of agglomerative (bottom-up, clumping) hierarchical clustering.

The figure is a diagrammatic representation of the process of agglomerative hierarchical clustering. Each dot represents a sample associated with a certain gene expression pattern. The computer algorithm will identify pairs of samples with the most similarity in expression pattern. Samples are added to these original pairs based now on the average pattern of gene expression of each of the pairs. This process is repeated, giving rise to larger and larger clusters and eventually all the samples will be grouped together under one cluster. The cluster can be represented in the form of a dendrogram as illustrated in Figure 8 (color page 545).