The visualization of a determined set of genes, whose expression varies across different pairwise experimental comparisons and whose corresponding proteins undergo a certain number of interactions, allows the analysis of microarray data when it is organized in dynamic network maps. Here, nodes are genes (or proteins), which are first connected by edges based on available protein-protein interaction information (PPIs, black lines). The network connectivity is further enriched by the addition of extra edges, which indicate that the expression of linked genes is correlated across all given experimental conditions (grey dashed lines). Finally, the layout of the network is organized based on their intracellular localization and their level of connectivity (spring layout). Nodes in the network maps are colored with a red-to-green gradient according to their expression values, along all the analyzed pairwise comparisons. Two different patterns (A and B) of gene expression behaviors could emerge by moving between the different network panels (x, y, z). This can be greatly facilitated by generating a animation from these network panels. Detected patterns can suggest biological interpretations and new hypothesis.
Visualization constitutes in itself a challenge. This is particularly true for the multivariate type of data that omics approaches generate. It is increasingly recognized that visualizing data is more than just presenting it; it also constitutes an exploration tool where the characteristics of human pattern recognition (Gestalt principles of perception) make the human subject a powerful ally of mathematical algorithms for the discovery of principles in complex data.