Representation of the PANA algorithm. Given a database of gene annotations into pathways and a gene expression experiment, an expression submatrix is generated per pathway by collecting the expression profiles of the genes annotated to each pathway. By applying PCA and bootstrapping to the expression submatrices, the Pathway Level Matrix is computed where row represent Pathway Profile. Next, a list of association rules between the pathway profiles is inferred by using GRNCOP. Finally, the pathway network is generated by selecting the rules with a high level of accuracy.