Cores in Chemical Reaction Networks From Dynamical Properties to Graph-Theoretic Conditions
- Peter Stadler (Leipzig University, Bio Informatics)
Abstract
The topological structure of chemical reaction networks places severe constraints on the possible dynamical behavior. We consider parameter-rich monotone reaction kinetics. In this setting, one can choose the numerical values of the positive entries "reactivity matrix" at will, while the zero-entries are defined completely by the structure of the reaction network. We then derive sufficient conditions for the existence of dynamical instabilities and periodic orbits. The key concept of "cores" emerges as well-defined minimal substructures defined by square submatrices of the stochiometric matrix that convey certain spectral properties. Minimality in the case may refer to arbitrary square submatrices or the principal submatrices only -- defining two distinct types of cores. As a special case of unstable structures we explore autocatalysis (in the sense of Nghe) in more detail. The relevant submatrices are so-called child-selections that have a graph-theoretical interpretation as fluffle-subgraphs of the bipartite representation of the reaction networks. Fluffels can be constructed as restricted superpositions of directed circuits. This provides an algorithmic starting point to identifying various types of cores in large networks. Autocatalytic cores are particularly easy to detect since their underlying fluffles form induced subgraphs of the reaction network. As a practical application I will illustrate autocatalytic cores in full-size metabolic reaction networks.