Klein, Brennan; Anshuman Swain; Travis Byrum; Samuel V. Scarpino and William F. Fagan
Understanding noise in networks and finding the right scale to represent a system are important problems in network biology. Most research focuses on the raw, micro-scale network from data/simulations and seldom explores the scale dependence of properties. Here, we introduce the einet package, which looks at the most informative scale in a biological network using recent concepts from information theory and network science. einet uses two metrics: Effective information, which measures the interplay between degeneracy and determinism in a network's edges, and causal emergence, which finds the scale of the network with the highest effective information. einet is available in R and Python and provides tools to explore noise and scale dependency in networks as well as compare information flow and noise across networks.