Fig. 9

Fitness function: maximizing versus minimizing number of edges. Comparison of predictive power (s _all score) of model networks obtained with the fitness function F = (f _T , − N _ess. nodes , N _nodes , − N _edges ) which minimize number of nodes and subsequently maximize number of edges (blue boxplots), and fitness function F = (f _T , − N _ess. nodes , N _edges ) which minimize number of edges (red boxplots). Except for s _all scores obtained by minimizing the number of edges (red), all results were taken from Fig. 8. Model networks optimized with minimization of the number of edges (red) used the same input data sets and the optimization procedure as in Fig. 8, except for the change of fitness function, and the use of 100 instead of 50 replicas