The Cell as a Machine

In the design of a robust device, it is common to specify all of the ancillary functions needed, and then determine how they interface with other functions in the physical device. The previous chapter discussed how to analyze a complex function in the cellular context by controlling cell phases, so that one can reproducibly analyze the same function in a known cellular context. In the design of a robust device, the designer determines what is needed; however, if the device (cell) was developed over many generations through robust selection processes, many ancillary functions or features may be hidden. The treatment of biological systems as robust devices should include an appreciation for all of the needed functions, the necessary links between them and the dependent parameters, such as ATP, which contribute to larger cellular activity. This complete treatment of the problem can enable one to analyze biological functions with a new perspective. The major difficulty in this approach is that we do not appreciate the complexity of most biological systems in that dietary changes, exercise levels, startle reflexes, and environmental factors such as temperature, bacteria, or viruses can all alter the normal balance of cellular homeostasis. In many of these cases, the organism has compensatory or adaptive mechanisms to minimize the trauma of an abrupt environmental change. This is part of the definition of a robust device. In this chapter, we will discuss how to dissect a primary function into a series of dependent functions and their governing parameters. Although many of those parameters are automatically controlled in cells, knowing that they are important may help to explain why certain perturbations cause unexpected changes in a given function.

‘Systems Biology’ has been defined as the study of how interactions between specific components of biological systems give rise to that system's function and behavior. For example, the proteins and the cell phase in clathrin-mediated endocytosis. Operationally, many have approached these questions by using protein expression, and interactomics data to generate models with a number of experimentally determined reaction constants. The problem that occurs in many cases is that the system is too poorly constrained and therefore the models are built with too many adjustable parameters. A rule of thumb in mathematical modelling is that you can model the shape of an elephant with four adjustable parameters and get it to walk with a fifth.