A minimal fitness model for evolutionary predictions

Predictions of future evolutionary processes have recently been developed for a number of systems, including the fast-evolving pathogens influenza and HIV. Two key molecular phenotypes have emerged as informative for predictions: protein folding stability and antigenicity, which is determined by interactions with the host’s immune system. A minimal fitness model based on these phenotypes shows time-dependence due to the changing pathogen environment generated by adaptive host immunity. I will show how this model can be used for predictions of antigenic evolution, how the relevant phenotypes can be learned from time-resolved sequence data, and how successful predictions feed back on our understanding of the underlying cell biology. I will use influenza virus and data from cancer patient cohorts as examples. Using these case studies, I will also discuss what are the emerging concepts for predictive analysis of fast-evolving systems.