An imperative language of self-modifying graphs for biological systems

The main design features of a language for the management of the dynamic reconfiguration of graphs are described. The language is domain-specific to model the behaviour of biological systems. Nodes of graphs are biochemical components, and undirected edges between them represent biochemical bonds. Also, nodes have a finite number of binding sites, and each of them can be the end-point of an edge leading to (exactly one end-point of) another node. A pair of nodes can be connected by multiple edges, each of them representing a particular bond between the involved biochemical entities. Spontaneous events, corresponding to the biological formation/breakage of bonds, can cause graph reconfigurations. Nodes can be programmed to react to them, and consequently change their own propensity to be involved in further events. The event handling routines coded within nodes are executed in a concurrent fashion. This poses the usual problems related to race conditions, deadlock and the like. Language primitives are tailored to cope with these issues. They ensure, e.g., consistency of reconfigurations: each binding site of every node can be connected at most with one single binding site of another node. This extended abstract describes the rationale behind the peculiarities of the language by showing excerpts of programmed nodes that, together with spontaneous events, can drive relevant reconfigurations of graphs.