Using AI to accelerate discovery and stockpile knowledge
New artificial intelligence tools are creating unprecedented opportunities to enhance fundamental virology research, allowing us to build the invaluable knowledge needed to tackle emerging viruses.
Researchers in the CVR have pioneered an approach using machine learning-enabled protein structure prediction to comprehensively map glycoproteins across the Flaviviridae family, which includes the hepatitis C virus (HCV), Dengue and Zika. This provided new evolutionary and genomic-scale perspectives of the entire family, revealing key molecular signatures that define the diverse virology and ecology found within the Flaviviridae. Building on this, the team combined two state-of-the-art approaches, AlphaFold2-ColabFold and ESMFold, to predict models for 85,000 proteins from 4,400 human and animal viruses, expanding the structural coverage for viral proteins by 30-fold compared to experimental structures. To enable further discoveries, we have created Viro3D (https://viro3d.cvr.gla.ac.uk/), a virus species-centred protein structure database. It allows users to search, browse and download protein models from a virus of interest and explore similar structures present in other virus species. This resource holds transformative potential for accelerating the discovery of vaccines and antiviral targets against existing and emerging viral threats.
AI is transforming virology and reshaping pandemic preparedness
Artificial intelligence is opening up entirely new ways of understanding viruses — and transforming how we prepare for future pandemics. In this video, researchers at the CVR explore how cutting-edge AI tools are revolutionising virology, from predicting protein structures to uncovering hidden viral diversity. Discover how these approaches are revealing new viral mechanisms, reshaping virus classification, and accelerating the search for diagnostics, vaccines and treatments.
Research Outputs
- Phenotypic evolution of SARS-CoV-2 spike during the COVID-19 pandemic
- PLM-interact: extending protein language models to predict protein-protein interactions
- Mapping glycoprotein structure reveals Flaviviridae evolutionary history
- Pathogen genomic surveillance and the AI revolution
- Viro3D: a comprehensive database of virus protein structure predictions
Resources
Viro3D is a comprehensive database of predicted protein structures from over 4,400 viruses, generated using state-of-the-art machine learning. By vastly expanding viral structural coverage, Viro3D enables exploration of viral evolution, form and function, supporting molecular virology research and structure-informed vaccine and therapeutic design.
Policy Contribution
Leveraging our expertise and leadership in this area, the CVR is collaborating with national government and international partners to inform and shape policy on the responsible application of AI in pandemic preparedness. Our aim is to establish frameworks that not only accelerate viral research but also provide a template for secure and innovative AI adoption across the wider life sciences.