Developments in Computational Models—Computational Models From Nature EPTCS 9, 2009, pp. 65–72, doi:10.4204/EPTCS.9.8 Rule-based modelling and tunable resolution

We investigate the use of an extension of rule-based modelling for cellular signalling to create a structured space of model variants. This enables the incremental development of rule sets that start from simple mechanisms and which, by a gradual increase in agent and rule resolution, evolve into more detailed descriptions. 1 Rule-based modelling Cell signalling networks involve many binding/unbinding interactions combined with post-translational modifications (PTMs) between a large number of primarily proteins. As is by now well-known [1], this gives rise to a combinatorial explosion in the number of possible molecular species and so in the number of reactions required to define a model. For example, a protein with ten phosphorylation sites already has, in isolation, 210 possible states which would mean 1024 distinct molecular species. A protein-protein interaction (PPI) depending on just one of those sites should therefore be independent of the states of the other nine; this would require 29 reactions, one for each possible state of those nine sites. This problem can be solved by using rules that act upon partially specified molecular species; so a rule determines an entire family of reactions. In our example, this replaces the 29 reactions with a single rule that mentions only the site of interest (and the other nine sites not at all). Rule-based modellin