Figure 1

Principle of the searching method for Best Scoring Clusters I. Figure parts show the work flow of the search method. As an input, the method takes the hierarchical cluster tree and calculated correlation values for each cluster in the cluster tree and each functional class (red and blue). II. Cluster tree from part I is shown with colour visualization referring to the most enriched gene class (blue or red) in each cluster based on the table shown in the previous part. Colouring is done to a U-shaped profile created by two joined clusters that create each cluster. Here the best result is obtained from the cluster F marked BSC with the blue class. III. Black colour presents clusters that will be silenced (omitted from search) after BSC is located. Note that the small red cluster is not silenced as the blue functional class is not the best enriched functional class and that the blue class is not silenced from other parts of the cluster tree. After silencing, the results from clusters B, G and I will not be taken into consideration any more and so the next BSC will be cluster D.