model making

A recent paper by Dalal et al: Rapid digital game creation for broadening participation in computing and fostering crucial thinking skills. International Journal of Social and Humanistic Computing, 2009; 1 (2): 123 investigated how the process of creating video games can boost students’ critical and creative thinking skills as well as broaden their participation in computing. The paper indicates that using rapid prototyping with visual representations and event driven behaviors develop creative thinking as well as computing skills.

Recently I have been going through the process of building a computer model of the gene circuitry I am building in the lab. I have found this activity to be quite beneficial, not only in the simulation tool I am creating but the process of creating it. I have found that simply thinking about the interactions of the various components and drawing out visual representations on paper, even without any sophisticated computer software helps to build on the strategy being undertaken in the lab. The approach highlights missing pieces of data and potentially important experiments to characterize the circuits behavior.

For the programming/modeling work (after a long time hammering away at Mathematica and Matlab) I decided to use the Copasi simulation and analysis software package as this gives me the fastest route to prototyping the gene circuit and provides the option to output as sbml which could later be implemented in Matlab if required. Copasi has enabled me to jump straight into the modeling from the point of view of the reactions and at the moment mass action kinetics seems to be sufficient to represent the genetic interactions. Copasi also enables optimization of the model. Signal to noise ratio is key to the circuitry I am building so I am able to optimize the model parameters with the input and output as objective functions. Copasi can then implement a number of deterministic and stochastic optimization algorithms to evolve the system towards the favored output.

The results of this approach have provided some hypothesis about the properties of the gene circuit components in order to achieve the desired output. This has provided a sort of first round iterative design/development cycle and once the lab data has been obtained should provide some interesting validation of the modeling/wet lab approach to the project.

More information on optimization algorithms can be found in Mendes & Kell (1998) Bioinformatics, Vol 14, 869-883.