Dr. Frank Tobin
GlaxoSmithKline
Some Computational Problems/Opportunities in Modeling Large Biological Networks
Modern biology is capable of generating finer and finer detail about biochemical and genetic interactions and there is an increasing need to integrate this information into mathematical models. This talk will focus on kinetic models of such networks - usually systems of ODEs. As the networks grow in size, the conceptual challenges remain relatively constant, but the practical aspects of model building, calibration (inverse problems), and model analyses progressively become unmanageable as the resulting dimensionalities increase. For example, the inverse problems in parameter estimation may require a global minimum in extremely high dimensions with an objective function containing multiple noisy datasets and non-linear constraints or penalties. With such problems, there are opportunities for new approaches to better understand, analyze and manipulate such models.