Broadly neutralizing antibodies (bnAbs) for highly mutable pathogens such as HIV and influenza can prevent and eliminate incurable diseases. We employ a computational approach that combines molecular simulation, bioinformatic analysis, and biophysical modeling of the adaptive immune system to design vaccines against such pathogens. Our work is directed toward a general and detailed model of affinity maturation to guide the choice of specific immunogens and immunization strategies needed to induce a broad and potent immune response. Previous work performed in collaboration with our colleagues at MIT used a coarse-grained model to provide proof-of-concept that the probability of emergence of bnAbs can be predicted. Building on this foundation, we use all-atom simulations and structural modeling to reduce the gap between extreme coarse graining and biological reality, to study real immunogens.