Difficult times in predicting flu evolution suggested by recent paper

There is a potentially controversial and very interesting article in the journal PLoS Pathogens on Flu Evolution. The study was led by Edward Holmes at Penn State, and co-authored by many researchers including colleagues of mine at my former institution TIGR. They performed a detailed evolutionary analysis of the cmplete genomes of 413 influenza A viruses of the H3N2 type (the H#N# system refers to the subtypes of Hemaglutanin and Neuraminidase genes).

The virus genomes were sequenced at TIGR using a high throughput flu viral genome sequencing protocol originally developed at described by Elodie Ghedin and colleagues here and here. The viruses they selected were from across New York State as part of a surveillance program.

Using a variety of evolutionary analyses including phylogenetic reconstructions and examination of substitution patterns, they come to a surprising conclusion – that

stochastic processes are more important in influenza virus evolution than previously thought, generating substantial genetic diversity in the short term

This may seem somewhat uninteresting to many out there but if true it is critically important in fighting flu and in understanding viral pathogen evolution. Right now there are substantial efforts to try and predict what future dominant flu strains will look like. These predictions tend to rely on assumptions that positive selection of viruses is critical in generating and maintaining diversity. If stochastic processes are as important as Holmes et al conclude, it would mean that more intensive monitoring of flu is needed in almost real time (since predicting random events tends to be, well, very hard).

I confess I have not tried to evaluate whether or not I think their conclusions are correct, but on first glance they seem sound. This just goes to show that general genomic surveys that try to be relatively unbiased in their sampling can reveal substantial novel patterns not seen before in highly target genome sequencing projects.