Since we want to understand the range of possible phenotypic variations in different animals' development we have decided to use modelling approaches. Our models take as inputs these gene networks and the initial distribution of cells in space (in a given stage in development) and provide as a result the final organ morphology and patterns of gene expression in a given organ (in a given, latter, stage of development).
Each model is simply a mathematical implementation of a hypothesis about how an organ develops. We construct these hypotheses, based on experimental work from other groups, and implement them in a computational model. The advantage of computational models in respect to merely verbal arguments is that the models provide precise quantitative predictions that are more easily to unambigously compare with experimental results (from new experiments aimed at testing the hypothesis). Merely verbal arguments are more difficult to be proven wrong or right and get even difficult to express when the process under study involves a large number of cells in complex movement and communication between them (as it is often the case in development).These easily lead to largely unintuitive dynamics that are hard to analyze without quantitative models. In addition, computational models allow to explore not only the wild-type but also, by variaton in the underlying gene network, the range of possible morphological variants (and how they change through development). The capacity to play with the parameters of the model allows us to actually understand its dynamics. Ultimately, a model is simply a summary of what we think we understand about a system but that allows us to see if the underlying hypothesis could work. That the model works does not imply that the hypothesis is right, further experiments are required, but if the model can not produce the right wild-type it means that the underlying hypothesis is wrong or incomplete. In other words, what we thought we understood, we did not actually understand.