Using Multi-Agent Systems to Design Internal Movements of Proteins
Résumé
Multi-Agent Systems (MAS) are systems composed of multiple autonomous entities (agents) which can collaborate. Applied to biological issues, they were used to model ecological systems such as insect colonies where agents can easily model the simple relationships between entities and the complex behaviour of the entire system. In the field of cellular and molecular biology, an agent is often a cell but rarely a single molecule. Moreover, these kinds of agents are not influenced by their shape or their physical features as they generally are non-physical agents. To model biological processes at the molecular level, especially enzymatic reactions, we have to take into account physical properties and explicit three-dimensional representation of molecules. Thus, we define a type of reactive agents composed of rules which correspond to the behaviour of agents to model enzymatic reactions and of a body which consists in the three-dimensionnal representation and the rules for physical interactions and internal movements. Three-dimensionnal structure are given by structural data extracted from the Protein DataBank (PDB). To determine possible internal movements, we analyzed these data with hinge determination algorithms and elastic network models to identify rigid parts of macromolecules and thus the possible conformational changes. The first application to respiratory chain complexes allows us to split these macromolecules in few parts (3 or 4) and we obtained a simple yet realistic representation.