Knowledge representation in systems biology

If biological data are being produced at an exponential rate, the production of models of cell signaling is lagging. One of the reasons for the unequal race between data production and data integration is the difficulty to provide large and combinatorial models with a proper semantics.

Site (or port) graph rewriting techniques, also called rule-based modeling [1], provide an efficient representation formalism to model protein-protein interactions in the context of cell-signaling. In these approaches, a cell state is abstracted as a graph, the nodes of which correspond to elementary molecular agents (typically proteins). Edges of site graphs connect nodes through named sites (sometimes called ports) which denote a physical contacts between agents. Biological mechanisms of action are interpreted as rewriting rules given as pairs of (site) graphs patterns. Importantly, rules are applied following a stochastic strategy, also known as SSA or Gillespie's algorithm for rule-based formalisms [2]. KaSim [3] is an efficient rule-based simulators that implements this algorithm for the Kappa language, a site-graph formalism adapted to biology.

In this talk, we will see how to abstract away and represent bio-molecular interactions in Kappa. More generally we will discuss the issue of Knowledge representation in the context of systems biology.

[1] Vincent Danos, Jérôme Feret, Walter Fontana, Russell Harmer, Jean Krivine: Rule-Based Modelling of Cellular Signalling. CONCUR 2007: 17-41 2006

[2] Vincent Danos, Jérôme Feret, Walter Fontana, Jean Krivine: Scalable Simulation of Cellular Signaling Networks. APLAS 2007: 139-157

[3] see dev.executableknowledge.org