Benchmarking of module detection performance with iMod and the Greedy algorithm. Synthetic network examples (128 nodes, 4 modules) were generated with node degrees of 5 and 16 in (a) and (b) respectively. For each mixing parameter, μ, 100 networks were assessed. The agreement of modules detected with the known community structure was expressed via the mutual information measure. Consistently better performance was noted for iMod in all examples tested.