AI created super vaccine against bird flu

By having AI analyze amounts of genetic sequences and then piece together a so-called antigen, British researchers have succeeded in creating a vaccine that is effective against different variants of bird flu.

- This new generation of vaccines is a "game changer". This means that we can be prepared and respond immediately in the event of an epidemic or pandemic, compared to today when we are always one step behind a virus strain that changes, says one of the researchers behind the study, Professor Jonathan Heeney at the University of Cambridge, in a comment.

Vaccines have saved the lives of millions of people. An obvious disadvantage, however, is that the vaccine manufacturers are often one step behind. Especially when it comes to viruses that change easily. The researchers simply cannot keep up, and in the meantime people risk becoming seriously ill and dying. Two clear examples are the vaccine against regular seasonal flu, which under great time pressure must be updated before each season, or sars-cov-2, which causes covid-19, with all its versions that appeared during the pandemic.

The dream would therefore be a vaccine that prepares the immune system against all conceivable variants of a certain virus, much like broad-spectrum antibiotics that kill a large number of different types of bacteria.

The most critical component of a vaccine is the so-called antigen. That is, the molecule, perhaps a piece of one of the surface proteins of the virus, that you want the immune system to react to, so that it is ready when the body is exposed to the real infection. How the antigen is designed is therefore key to how the vaccine will work. Preferably not only against the virus variant that is currently circulating, but also against those that could arise if the virus changes.

One of the viruses that scientists and authorities fear the most is bird flu, of which there are many variants. The virus is mainly found in birds, but it happens that humans are infected. The nightmare would be a new virus variant with high mortality and the ability to infect people.

The researchers in Cambridge started from the genetic sequences of all the bird flu variants identified by surveillance authorities worldwide. Then they had an AI tool analyze the sequences and suggest a number of “super-antigens,” after which the researchers used cells to screen the best candidates for inclusion in a vaccine.

The vaccine was then tested in mice and ferrets that were exposed to different variants of bird flu and compared to one of the candidate vaccines included in the WHO's work with pandemic preparedness. Ferrets are common in this type of research, as they react to flu viruses much like humans do. The results, now presented in Cell Reports Medicine, show that in the group that received one of the WHO's vaccines, about half of the animals died, while all the animals survived among those that received the new vaccine in a lower dose. In the control group, which received no vaccine at all, all animals died.

Vaccine expert Ali Mirazimi, adjunct professor at the Karolinska Institute, definitely believes that AI can become a valuable tool for developing new vaccines. However, completely depending on which infection it is about.

- AI speeds up and simplifies processes, which is useful when there are very large amounts of data, i.e. information. Some viruses, such as influenza, we know a great deal about. There are large amounts of data and then AI can clearly be useful. We know much less about other viruses and then AI will not be as useful, he tells DN.

Is it possible to make a vaccine that works against many different variants?

- Over the years, many have tried, but no one has succeeded. Here, AI could be useful in finding the lowest common denominator, so to speak. On the other hand, we don't know what these new variants might look like. It might be a variant that binds to the protein in a completely different way. In that case, the vaccine may work less well or not at all.