Statistical and computational challenges for whole cell modelling

Mathematical modelling of whole biological cells opens up new opportunities for fundamental and applied biology. In particular in the context of synthetic biology, it opens up the scope for rational engineering and design principles to be applied. But there are precious few such models available. Here I outline the challenges in the way of generating such whole cell models. The inference of parameters, the choice among competing models, and, first and foremost, the reliable construction of such models pose considerable challenges. Recent work in statistical inference, especially parameter estimation, and model selection, coupled to new computationally more efficient methods to simulate large (and stochastic) biochemical reaction systems will be pivotal for the generation of a new generation of whole cell models. But these need to be coupled to better ways of generating models de novo. I outline how this may be achieved, and why this is necessary.