Congratulations to Gemma Atkinson, Roni Odai and Vasili Hauryliuk at the Department of Experimental Medical Science, Lund University, and their collaborators, on their study “The Viral AlphaFold Database of monomers and homodimers reveals conserved protein folds in viruses of bacteria, archaea, and eukaryotes” that is published in the high impact scientific journal Science Advances. This study contributes valuable insights to our basic scientific understanding while also offering important implications for future pandemic preparedness.
Recent advances in AI methodology have revolutionised bioinformatics. AlphaFold is an advanced AI method for protein structure prediction.
“It means we can get a confident atomic-level structure within seconds, instead of having to purify and crystallize a protein to get an experimental structure. It has revolutionized biology as a whole” says Gemma.
Since the publicly available AlphaFold database of predicted structures lacked viral proteins, it inspired the Atkinson lab to build the Viral AlphaFold Database (VAD), based on nearly 27,000 high-quality viral protein models. This database opens up new possibilities for exploring viral protein structure, function and evolution.
“We also included an angle that is quite unique for our project”, Roni continues, “which is to assign a host type to each protein and from there find what protein structural folds are reused in viral infections of different domains of life. It’s fascinating to see if the folds are really reused or if some are specific to viruses infecting humans or bacteria- and it turns out both are true!”
The study sheds light on the evolutionary relationships between viruses that infect humans and those that infect entirely different organisms, such as bacteria, offering new insight into their evolution, infection strategies, and defense mechanisms. Gemma explains:
“It wasn't possible to see this before because viruses evolve so quickly that you don't see the similarity at the sequence level. By comparing structures, we can find more distant relationships that you can see with sequences alone. We now know a bit more about the evolutionary relationships of proteins in viruses that infect humans and other eukaryotes and also those very different viruses that infect bacteria- the bacteriophages. This was a big motivating factor for conducting the study.”
When asked if anything unexpected emerged from the study, Roni replies: “Yes! A lot of the protein folds are reused across infections of different domains of life. We weren't expecting there to be that many, particularly if you look at the protein sequence alone. But on the structural level, this number increases dramatically. Another thing we found that was surprising was that many proteins related to pathogen- host-interactions are among these what we call hetero-host folds.”
The VAD ViralAlphaFold Database provides a foundation for discovering antiviral defense systems, drug targets, and biotechnology applications, and underpins ongoing work in the Atkinson Lab to focus predictions on key viral lineages — increasing protein knowledge for future pandemic preparedness.
“This paper is quite broad across many different kinds of viruses, which allow us to ask the big questions about relationships among very different viruses. There is still much to be discovered about the proteins in families of viruses that could be the source of the next pandemic. This technology will definitely be used in future pandemic preparedness- especially in the next phase of the project where we're focusing more on human infecting viruses” says Gemma.
The ViralAlphaFold Database is publicly available, offering a valuable resource for researchers worldwide.
“We have a lot more information about what viral proteins look like,” says Roni. “If you have a protein and you don't know what it does, you can use our database to hopefully find something that is structurally similar and then maybe even find some clues about what the function of that protein is. It’s just a lot more information than what was previously available when you only have a sequence of letters- now we have 3D coordinates that you can leverage.”
Gemma is the director of LU-Fold, a SciLifeLab infrastructure to help other researchers use AlphaFold in high-throughput analyses. More information on LU-Fold can be found at https://www.medicine.lu.se/research/list-research-infrastructures/lu-fo….
Read the full article The Viral AlphaFold Database of monomers and homodimers reveals conserved protein folds in viruses of bacteria, archaea, and eukaryotes in Science Advances
Corresponding authors are besides from Gemma Atkinson and Roni Odai, Vasili Hauryliuk at Lund University, Janani Durairaj at University of Basel, and Joana Pereira at Leuven University.
Gemma Atkinson and Vasili Hauryliuk are affiliated with the Lund University Virus Centre (LUVC).