Finding the appropriate algorithm is essential to the automatic design of CSNs. Given the plethora of available approaches and the lack of experience and knowledge in the evolution of biological networks, it is not straightforward to create an efficient algorithmic scheme. The ESIGNET project is currently generating vital progress and experience in this field, and new insights will certainly be gained from the combination of well-proven approaches into a software capable of the artificial evolution of CSNs.
Considering the number of algorithmic combinations to choose from, it is of vital importance to design any approach in a modular way, so that individual parts of the algorithms can be exchanged and the results can be compared. Results from these comparisons will have scientific value of their own, as well as giving hints to the further development of CSN-evolution software. In the discussion above, we gave of short insight into the diversity of design principles that are currently used and outlined a possible combination of suitable methods for our project. As a promising example, we have chosen a graph-based Genetic Programming method, combined either with direct numeric mutations or an ES algorithm. First promising results will be reported on in other workpackages.