In addition, only a handful of the epitopes are known to be associated with protective immunity. Most of the influenza virus epitopes in the database are those recognized by a type of immune cell known as a T cell; far fewer are recognized by B cells, a type of white blood cell that produces infection-fighting antibodies. Antibodies induced by seasonal and pandemic flu viruses or vaccines are a major component of immunity that protects against these viruses.

Strains of influenza virus can vary enough in their neutralizing B cell epitopes that a vaccine against one strain may not protect against another strain. But if epitopes are conserved between virus strains, the immunity a person has developed towards one strain might provide at least some protection against the other strain.

Using a software tool they developed, the LIAI team found hundreds of conserved influenza virus epitopes in the database, including those between avian H5N1 and strains of human influenza viruses. But what is less clear from the analysis is how cross-reactive an immune response would be to most of these conserved epitopes. Further analyses may assist scientists in identifying vaccine targets that might offer broader protection and in predicting how effective a new vaccine will be.

Other analyses revealed major gaps in scientists' knowledge about influenza viruses. Of the 600 epitopes in the database, for instance, very few were from strains of H5N1 avian influenza. And even though the database contains epitopes from all the influenza virus? proteins, the vast majority of the data relates to just two influenza proteins, the hemagglutinin (HA) and nucleoprotein (NP).

Most of the influenza virus data comes from analyses of immune responses obtained with mice; some comes from rabbits, ferrets and monkeys, and very little comes from humans or birds. In fact, only one antibody epitope came from a human. The LIAI researchers say more studies should be focused on identifying human T and B cell epitopes from human and avian strains of influenza virus — especially those associated with protective immunity.

"The bottom line is that this study shows us where we need to go," says project director Stephen Wilson, Ph.D., chief technology officer at LIAI. "Hundreds of flu epitopes have already been published and are now in the database, but critical gaps become apparent when one looks for human antibody targets."

Plans for the future include adding data on epitopes that are involved in autoimmune diseases and epitopes that trigger allergic and asthmatic reactions. Dr. Sette and his colleagues have also built numerous tools for analyzing and visualizing the data and for predicting immunity against different pathogens — all of which is publicly accessible on their website (see http://immuneepitope.org).