Elements of Computational Systems Biology

4

RULE BASED MODELING AND MODEL REFINEMENT

Elaine Murphy and Vincent Danos

University of Edinburgh, Edinburgh, UK

Jérôme Féret

ENS-INRIA-CNRS, Paris, France

Jean Krivine

Harvard University, Boston, Massachusetts, USA

Russell Harmer

CNRS-Paris-Diderot, Paris, France

Rule-based modeling is an effective way of handling the explosive combinatorics of biological networks. The use of partial objects in describing molecular interactions means that only the necessary conditions for a rule are specified and not the complete chemical entities taking part in a reaction. This leads to descriptions that are easier to set up and more compact. Networks of substantial scale can be described without having to reduce the combinatorics of the system—as other approaches must.

An important aspect of the rule-based approach is its agility, as one can easily modify rules to incorporate new knowledge or test different assumptions. A special and rather frequent case is when one wishes to replace a rule with ones imposing stronger conditions. This process is called refinement, and we approach it in this study both from the practical and the theoretical point of view.

There are various reasons why one would like to use refinement:

• One wants to understand how the activity of a rule varies with its application contexts
• One realizes that more conditions are necessary than previously thought
• One more subtly wishes to evolve the behavior of the current system

The notion of behavior-preserving, or neutral, ...