Last year I described the arm's race happening between virus and immune system during an HIV infection: as the immune system starts mounting its defense against the virus, the virus mutates trying to evade the attack. This is what pushes the virus to constantly evolve new strains, not just in HIV, but also in the flu virus, which evolves a new strain roughly every year. The HIV virus evolves within the same host to evade the host's immune response. On the other hand, the flu virus evolves more slowly: contrary to HIV, healthy individuals can clear the flu virus, and in doing so they acquire immunity against future infections from the same virus. This exerts pressure on the flu virus to evolve new strains capable of evading the population acquired immunity.
The process by which viruses constantly evolve new strains in order to evade immune responses is called antigenic drift.
The yearly evolution of the flu virus is closely monitored: surveillance data is collected throughout the flu season and, based on the data, a prediction is made on which strains will be most likely to reappear during the next season -- this step is important for flu vaccine design. The vaccine needs to be available prior to the start of the new flu season. Therefore, researchers have to make an educated guess on what the evolved flu virus will be like in order to make the appropriate vaccine.
"Due to the fast evolution of the influenza virus, the components of the influenza vaccine are changed for many flu seasons. Even though the vaccine is usually redesigned to match closely the newly evolved influenza virus strains, there occasionally has been a suboptimal match between vaccine and virus ."The surveillance data comes from the World Health Organization Global Influenza Surveillance Network (GISN), a network of 136 national influenza centers scattered in 106 different countries. The data focuses on one influenza gene in particular, the hemagglutinin (HA) because the protein it codes seems to drive the antibody response.
The HA protein coats the outer surface of the influenza virus. It enables the virus to recognize and bind target cells. Once bound to the surface of the cell, the virus is engulfed in a sac called endosome. This is a mechanism by which cells engulf extraneous objects and then try to destroy (digest them through enzymes) while inside the endosome. However, the influenza virus uses the endosome to get inside the cell and once there the HA protein undergoes a conformational change (triggered by a drop of pH) and becomes a "hook" that breaks the endosome and frees the virus into the cytoplasm. Without the HA protein the flu virus is unable to bind to the target cell or break the endosome. Therefore, antibodies that bind to the HA protein successfully clear the virus, which is why it is vital for the virus to evolve mutations that enable it to escape those antibodies.
In order to anticipate the next flu strains, researchers need to understand how well the population is responding to the current strains. The flu vaccine usually carries three different strains, selected from the most predominant and geographically spread ones so that the resulting immunity is reactive to a wide range of flu strains. How "different" any two strains are is measured by a quantity called "antigenic distance," which, in layman terms, measures how well current immune responses ("animal antisera raised against the same or related strains, ") are able to block those strains. Viruses with a high antigenic distance will be poorly blocked by the current immunological responses and therefore are more likely to diverge from the current flu strains and spread into the following season.
In , Smith et al. reconstruct the antigenic map of the influenza A virus starting from 1968. The map retraces the genetic evolution of the virus, showing that strains tend to form clusters that last 2-3 years and then evolve into a new cluster (a new strain that requires a new vaccine). New surveillance techniques are being developed based on this concept of antigenic distance and antigenic maps in order to help inform future vaccine selection.
 Pan K, Subieta KC, & Deem MW (2011). A novel sequence-based antigenic distance measure for H1N1, with application to vaccine effectiveness and the selection of vaccine strains. Protein engineering, design & selection : PEDS, 24 (3), 291-9 PMID: 21123189
 Cai Z, Zhang T, & Wan XF (2012). Antigenic distance measurements for seasonal influenza vaccine selection. Vaccine, 30 (2), 448-53 PMID: 22063385
 Smith DJ, Lapedes AS, de Jong JC, Bestebroer TM, Rimmelzwaan GF, Osterhaus AD, & Fouchier RA (2004). Mapping the antigenic and genetic evolution of influenza virus. Science (New York, N.Y.), 305 (5682), 371-6 PMID: 15218094