Graphical Models for HIV Vaccine Design

I will discuss two applications of graphical models to HIV vaccine design. The first helps determine how strongly our immune system fights HIV. The second helps identify which parts of HIV can be successfully attacked by our immune system. I will also discuss how these applications have exposed a weakness in the process of learning graphical models from data---namely, the inability to quantify how many arcs in a learned graphical model are spurious. I will offer a solution based on the False Discovery Rate.