Induction of broadly neutralizing antibodies in Germinal Centre simulations.

Vaccines against mutating pathogens such as influenza, HIV, or plasmodium are poorly protective towards new evolving strains. Rare individuals naturally mount broadly neutralizing antibodies covering most strains, but the requirements for their induction are unknown. The antibody response to vaccination has been recapitulated by in silico models that proposed two opposite schemes: A theory of 'frustration' where one epitope at a time leads to optimal antibody breadth through sequential immunizations, that was proven successful for HIV vaccination in primates. Another theory supports vaccination with cocktails of multiple representative epitopes in a unique prime and boost, which succeeded for influenza in mice. We discuss how in silico models differ in their assumptions, with particular focus on protein affinity representation.