The Perceived Directionality of the Antigenic Evolution of Influenza

Abstract

The in influenza virus can evade the acquired long term immunity in the human population through changes in its antigenic profi?le. Understanding the antigenic evolution is crucial to the selection of vaccine strains, which should provide immunity against dominant influenza strain next season. For this purpose, antigenic cartography has been developed to map the changes of the virus over time, using Hemagglutination Inhibition (HI) assay data. Here we present a simple model for the antigenic evolution of the influenza virus, based on a high dimensional random walk in Euclidean space. Using this model we simulate HI assay data which we use to construct antigenic maps through antigenic cartography. Our analysis shows that, although the clustered structure of the data is preserved, the number of dimensions required to describe the data is greatly underestimated. This means that the techniques used in antigenic cartography to assert the dimension required to describe the HI assay datasets does not give conclusive results about the actual dimension of the shape space in which the antigenic evolution takes place. The assumption that the anti- genic evolution of influenza is influenced by long term immunity in the human population can therefore not be veri?ed using HI assay data. Our analysis shows that directionless evolution through a high dimensional shape space forms a parsimonious explanation for the observed pattern of influenza evolution.