Systems biology's dirty secret: Parameter estimation, sensitivity analysis, and sloppiness

Models of complex biological phenomena often involve many quantitative parameters (such as rate constants) that are difficult to measure directly. Consequently, such parameters are often fit to more complex data. I will first discuss various approaches for this often challenging problem of fitting model parameters to data. Fortunately, nonlinear models of complex phenomena often exhibit a "sloppy" pattern of parameter sensitivities, with some combinations of parameters exponentially more sensitive than others. I will discuss evidence for the universality of such sloppiness and the implications of sloppiness for model fitting and experimental design.