Enumeration of all MCSs (synthetic reaction lethals) in a medium-scale metabolic model of E. coli.

Computation times for MCSs that disable growth in an E. coli metabolic network model of the central metabolism under different substrate uptake conditions (cf. Table 1 in [22]). The full/compressed networks contain 89/25 metabolites and 106/42 reactions. Conversion of MCSs in the compressed network to those in the full network takes negligible computation time for the cases shown here. For iterative solve (ALGO1) CPLEX dynamic search was used while for populate calls (ALGO2) traditional branch-and-cut was applied. In the fifth problem, only the MCSs up to size 4 were calculated. The computation times for the classical approach (EM+minimal hitting sets) and for the dual approach of Ballerstein et al. in the first four problems are the same as in [22]; note that neither method currently supports multiple threads. For calculations using multiple threads the physical computation time is shown with the sum of computation times (CPU times) over all threads in brackets. The calculations with 1 and 4 threads were performed with an Intel Q9550 desktop processor (4 cores) while for 12 threads a cluster node with two Intel Xeon DP X5650 processors (each 6 cores) was used.