Artificial Intelligence accelerates the path to new vaccines

Imagine a universal translator that, instead of turning English into Italian, can decode the language of the proteins that make up viruses. Such a “translator” already exists: Artificial Intelligence. And it is reshaping the fight against viral diseases, from pandemic preparedness to the development of treatments. This was the key message from international experts who, today in Trieste, outlined the latest frontiers of computational virology during the workshop “AI in Virology: Leveraging AI to Advance Our Understanding of Viruses”, hosted by the Virology Unit of Area Science Park.

For decades, the only way to study a virus was to grow it in the laboratory and observe its behaviour — a slow and costly process. Genetics then opened the door to reading its “instruction book”: the genome. Now AI goes further, learning the “grammar” and “syntax” of proteins — the molecular machines that allow a virus to invade cells and replicate.

“New language models for proteins are like artificial brains trained on millions of biological sequences,” explains Giuditta De Lorenzo, virologist at Area Science Park. “From a single sequence of amino acids, they can identify which mutations are possible and which would instead ‘break’ the protein. This makes it possible to predict how a newly discovered virus might evolve — a crucial skill if we want to stay ahead of future pandemics. For example, our upcoming research at Area will focus on the impact of viral infection on the cell: how viruses disrupt its contents. And in collaboration with our Data Engineering Laboratory, we will also work on developing vaccines that are more effective, more stable, and designed to take into account the dynamic behaviour of viral particles

Ultra-rapid vaccines thanks to “Reverse Vaccinology 3.0”

One of the most tangible impacts of AI will be on the development of vaccines and therapeutic antibodies. The so-called “Reverse Vaccinology 3.0” uses AI to instantly analyse the structure of viral proteins and identify their “Achilles’ heel” — the precise point at which antibodies can strike. “The huge advantage of Reverse Vaccinology 3.0,” explains Emanuele Andreano of the Biotecnopolo Foundation in Siena, “is the ability to discover antigens for vaccine candidates at unprecedented speed. Thanks to AI, and to advances in human immunology, it is now possible to quickly identify antibodies capable of killing a pathogen and then, from the antibody sequence, determine the target — the antigen on the surface of the virus or bacterium. This leap allows us to skip years of in vivo testing, understanding from the outset what works and what doesn’t. At the Biotecnopolo Foundation in Siena, our most important mission is to develop vaccines and monoclonal antibodies against viruses or bacteria with pandemic potential, such as the case of the monkeypox virus.”

However, as noted at the workshop, this immense computing power comes with very high costs. Behind these breakthroughs are supercomputers that consume enormous amounts of energy. The public must be aware that AI, while extremely powerful, is also very expensive and demands significant investment in infrastructure.

A promising future, but one to be governed with caution

The ability to read, interpret, and even “imagine” new proteins is not just an opportunity but also a profound responsibility, experts warned. “We must create shared international rules and robust control frameworks to ensure that this extraordinary scientific revolution is used solely for the benefit of humanity,” stresses Alessandro Marcello, virologist at ICGEB. “We have to consider the dual-use potential of AI, which can be very beneficial for medicine and public health, but could also pose risks if it fell into the wrong hands, given how relatively easy it could become to obtain protocols for producing highly pathogenic viruses. We must act synergistically on multiple levels: among AI developers, within the scientific community, and at the legislative level, to establish laws and regulations that, while not stifling research and innovation, protect society from these potential dangers”.