All fruitful collaborations require the collective agreement on the goal, defining approaches to attain that goal, and importantly, identifying the underlying assumptions (). This can be challenging, as fields are sometimes steeped in age-old ideas whose validity may not apply across the board or may violate some physical principles. Simultaneously, it is essential to critically assess what the available data really support. It is advantageous to become very rigorous and strict in the language used to describe any particular set of observations, as this can help close the language gap between the disciplines, helping to ensure that everyone develops a consensus view of what is known and what is not.
Diagram depicts the iterative workflow for developing and testing the physical underpinnings of a cellular process.
With this framework in place, one can begin to develop physical models and theories to explain a biological observation. Typically, models in cell biology begin as cartoon depictions, which are designed to summarize available data and frequently include molecular pathways. However, one ultimately wants to evolve these cartoon depictions into mathematical models (either analytical or computational) so the model can be tested against physical principles. One ideally wants to develop the model based on a subset of data, reserving other data sets (such as those from a different series of mutants that alter the parameters of the system) to challenge the predictions of the model.
Because diverse systems have different levels of biological complexity and may be better or less well understood, they may require different approaches for analysis and modeling. A poorly characterized system may not be ready for a modeling effort, or may only allow a simpler model that captures a few key aspects of the process. These simple “toy” models can still be quite useful, especially if they are based on measured (or measurable) variables, and the model output can be compared with an experimentally determined parameter. Regardless of the level of complexity appropriate for the model, some of the questions are the same. Are the assumptions right? Is the model physically and biologically grounded?
Models, whatever their form, serve several purposes. Preferably, they help motivate experiments by making testable predictions. However, even if this is not possible, models provide an independent test of our understanding—as simulations based on these models should recreate experimental results, while simultaneously following physical laws. Examples of how these laws manifest themselves in cell biology include the constraints on intracellular transport, the diffusion-based limits on reaction rates, and how forces can deform a viscoelastic network.