Information Theoretic Properties of Population Genetic Data
There has recently been renewed interest at borrowing both concepts and technical results from information theory for analysis in the biosciences. I will briefly review some efforts at incorporating notions such as entropy, channel capacity, communication noise, Kolmogorov complexity and mutual information into a biological framework, beyond merely as metaphorical. I will then describe a recent proposal that highlights intrinsic information-theoretic properties of genetic population samples. In essence, long stretches of genetic variants may be captured as messages generated by a nonstationary communication source modelled on a number of target populations. This perspective motivates the construction of simple population assignment schemes based on information theoretic quantities, where noise inherent to the learning process plays a similar role as channel noise in communication.