Multi-scale network properties and cartographic roles. The multi-scale network analysis involves exploring the network properties of the proteins at different scales, namely local, meso-scale, and global. The degree of a node is a local network property, since it considers the first direct neighbors of a node, while the shortest path is a global network property, since we need to count paths between pairs of nodes across the whole network. Finally, the meso-scale network properties represent the organization of the network into clusters or modules. The meso-scale connectivity features of each protein can be characterized with the cartographic role classification scheme proposed by Guimerà and Amaral, 2005, namely ultra-peripheral, peripheral, non-hub connector, non-hub kinless, provincial hubs, connector hubs and kinless hubs (see Supplementary Figure S1 for more information). Thus, focusing on how individual nodes are positioned in the modular (meso-scale) structure of the network, we can identify proteins that play different functions, such as mainly connected to other proteins within their modules (e.g., provincial hub), and those proteins that serve as bridges between modules (e.g., kinless hub).