Fig. 2

Illustration and analysis of D1, D2 and D3 network topologies. The redGEM algorithm was used to generate D1 (blue), D2 (red) and D3 (green) core enzymatic reaction networks composed of 271, 307 and 327 reactions, respectively. Core network metabolites that change in connectivity (a) between D1 and D2, and (b) between D2 and D3 are highlighted. These additional connections/reactions result in increased flexibility of the network. The schematic representation (c) of the studied metabolic networks shows the reactions (edges) and metabolites (nodes), and how they are connected via lumped reactions (dashed line) to biomass building blocks (brown ellipsoid). There are 102 biomass building blocks (listed in Additional file 3) in the E. coli iJO1366 that are preserved across reduced models. Reactions from D1 (blue) and D2 (red) correspond to the core of the metabolic models. The lumped reactions can be unique to D1 (blue) or D2 (red), or be common between both (black). Fluxomics data (black solid arrows) were integrated for optimally grown E. coli [17]. Each lumped reaction is composed of multiple reactions lumped together, also referred to as a subnetwork. Venn diagrams highlight differences in the lumped reactions of D1, D2 and D3 in terms of (d) subnetworks and in terms of (e) reactions composing them