Biology confronts us beautiful examples of evolved networked systems, where network nodes — be it genes, neurons, or whole organisms — interact to perform a certain function. Even though the individual nodes may be heterogenous, unreliable, delayed, or noisy, the emergent collective behaviors can be stunningly reproducible. One way to understand this evolved complexity is via the use of optimization principles, by assuming that natural selection has driven such networked systems far away from random expectation, or typicality. In this course, I will try to touch upon multiple intersecting themes where optimization principles help us rationalize and perhaps even predict, ab initio, what is observed, in particular, with reference to how living systems process information. I will:

-) introduce a Bayesian statistical framework that interpolates between ab initio prediction (via optimality) and data fitting

-) give a historical background and some recent results on how optimization of information transmission has predicted non-trivial things about our nervous systems

-) talk about information transmission in genetic regulatory networks and apply the ideas to early fruit fly development

-) show some very recent work that ties together evolution and what may enable or prevent information from being optimally encoded in our genomes

These lectures bring together ideas from statistical physics, information theory, evolution, and molecular biophysics, but will not assume extensive background in any of the topics.