Physics 434, 2012: Block three: Dynamical Information Processing
Back to the main Teaching page.
Back to Physics 434, 2012: Information Processing in Biology.
In this block, we started with pointing out that information processing in biological systems is not just probabilistic, but also dynamic, and we worked our way through analysis of dynamic information processing.
- We pointed out that two approximations help us in this analysis: quasi-steady-state approximation, and linear approximation
- We derived dynamical laws for a few biochemical systems in the quasi-steady-state approximation
- We introduces Fourier analysis technique that allows us to calculate signal transduction properties of various biochemical networks near equilibrium, where fluctuations areound the steady states are small, and responses can be linearized. This was done largely on the example of signal transduction in vertebrate retina.
- We introduces frequency dependent gain and phase shift as the means of characterizing signal transduction and studies how linear negative feedback affects the quantities.
- We realized that, once introduced, noise cannot be removed from the system, but computations can be done to make specific information available at maybe an earlier time. Typically, these calculations involve some kind of averaging: temporal, spatial, over a number of reactions, and so on.
- We further studied the idea of computation as trying to extract some relevant bits, but not all information, from the signal. A simple example of such extraction is thresholding.
- From there we did some dynamical systems theory to understand how such computations as thresholding and commitment to an outcome can emerge from simple dynamical equations. We then proceeded to study oscillations and emphasized that, due to various phases, separated by bifurcations, it's generally impossible to predict what a particular system does based on just its wiring diagram, but without knowing its parameters.
