Figure 4

The outline for clustering analysis of pathway activity levels. Pathway activity analysis begins with mapping gene expression onto known pre-defined groups of genes, pathways. Subsequently, the pathway activity levels are calculated using SVD and the significance of pathway activity levels are evaluated. Pathways are filtered based on the significance of the PALs. Following, the over-populated patterns are identified by using a consensus clustering approach proposed in [19]. Then, the parameters of the sinusoidal model A · sin(B · t + C) that would best fit the centroids of the pathway activity levels (in each clusters) are characterized. Finally, the correlation between fitted sinusoidal model and the centroids of the pathway activity levels in each cluster is evaluated.