ABSTRACT

Complex functional networks in biology have non random structures shaped by evolution. In this chapter, we review a general method based on evolutionary computation to explore the space of possible networks performing a specified biological function, using the example of immune ligand recognition as a case study. We reframe this problem using an information theory based “fitness", discuss the general solution found by our algorithm (that we name “adaptive sorting") and compare to actual biological networks. We finally establish the connection between ligand recognition and antagonism, the latter being a necessary undesired consequence of the former. Evolutionary computation thus constitutes a general predictive approach leading to experimental and theoretical insights on the “actual and the possible" in biology.