Hierarchical clustering with distance matrix

Hierarchical clustering builds a hierarchy of clusters. J-Express uses an agglomerative strategy for clustering. For this component a distance matrix is also presented. The distance between two objects can be measured by different distance measures. Colour codes are used instead of numbers in the matrix to give it a better visual representation. Since the genes are ordered according to their position in the hierarchical clustering tree, similar genes will appear in adjacent positions, making it easier to spot clusters within the distance matrix.

  1. Click the Clustering button ( Hierarchical Clustering With Distance Matrix ) button or select from the Methods | Clustering | Hierarchical clustering with distance matrix from the J-Express menu bar.
  2. Choose a distance measure and a linkage method. (Even if you use default values it may help interpreting the results if you understand how the methods work)
  3. You have to choose between clustering the Rows (Genes) or Columns (Samples). If you want to cluster the rows, it is advicable to only do this for relatively small datasets as hierarchical clustering is very memory consuming.