On the Semantic Capacity of (Bio-)Chemical Systems

The basic idea is to measure how easy it is to implement an organic molecular code with this network. Inspired by Barbieri (2008), we define a molecular organic code with respect to a given reaction network as a mapping between two sets of molecular species called signs and meanings, respectively, such that (a) this mapping can be realized by a third set of molecular species, the codemaker and (b) there exists alternative sets of molecular species, i.e., alternative codemakers, implying different mappings between the same two sets of signals and meanings (Görlich and Dittrich, in press).For an example see figure . We define the semantic capacity of a reaction network by simply counting the number of different codes. We analyzed models of real chemical systems (Martian atmosphere chemistry and various combustion chemistries), bio-chemical systems (gene expression, gene translation, and phosphorylation signaling cascades), as well as random networks and artificial chemistries. We found that different chemical systems posses different semantic capacities. Basically no semantic capacity was found in the atmosphere chemistry of Mars and all combustion chemistries, i.e., with these chemistries, organic codes cannot be implemented. Whereas the bio-chemical systems posses very high semantic capacities, with (hypothetically) increasing capacity from metabolic networks, signalling networks, to gene regulatory networks. andom networks have a much lower semantic capacity than biological networks like regulatory networks or the genetic code network. Random networks show only organic codes for very specific parameters, for example a random network with 15 species and an optimal density of reactions (i.e., 30) has on average 2.7 code pairs whereas a gene regulatory network of the same size has 9 code pairs. This can be explained by the fact that it is hard to achieve at the same time a high number of closures and a large pool of pathways to select from. Note that for a code pair at least ten different closed sets are necessary.