Bottom-up modelling in biology is quite different from the traditional mathematical approach. The technique uses software constructs or ‘agents’ as the computational equivalents of proteins and cells. These are programmed into a virtual finite state machine – an X-machine – that is Turing complete (it can compute anything). The number of agents can be increased to cover everything considered important to a problem, from molecules to populations, as long as enough computing power is available!
This approach has been used to considerable effect at Sheffield to model cell signalling, where receptors, biologically active molecules and membrane transport processes were each modelled by a software agent. Following the motions of the agents over time within the cell gives considerable insight into how signalling processes might work, and allows researchers to test hypotheses.
Nuclear factor κB (NF-κB) is a transcription factor induced through cellular signal transduction, and is central to inflammatory and immune responses. Modelling NF-κB pathway activation is essential in order to gain a better understanding of how it can be controlled, helping to overcome the practical limitations of biological experiments.
Methods previously used to model signalling pathways, such as reaction kinetics and Pi-calculus, consider pathways as being nothing more than the chemicals involved. Whilst this is appropriate in many circumstances, a more realistic description of the cell is necessary to understand the dynamics of the NF-κB pathway, where cell structure and mechanical stimulation play a vital role.
Agents are used to create a more realistic framework with which to model the intracellular NF-κB signalling pathway, incorporating explicit spatial dimensions and allowing the modelling of low numbers of non-uniformly distributed molecules.
Above right: A diagramatic model of the interactions between agents modelling the NF-κB signalling pathway. In this model individual molecules are each represented by a software agent.