Virus versus machine: using artificial intelligence to characterise viral entry mechanisms.
Supervisor: Dr Joe Grove, School of Infection & Immunity
Video introduction here.
Overview: Machine-learning (ML) is revolutionising molecular biology, enabling the discovery and investigation of novel molecular mechanisms, and paving the way to innovative new treatments for disease. We have pioneered the use of this technology to investigate cell entry mechanisms in hepatitis C virus. This has resulted in immediate fundamental discoveries that will guide our research programme for years to come and potentially inform vaccinology. However, viruses are absent from much of the training data used to generate state-of-the-art ML tools such as AlphaFold and ESMFold, therefore, these tools remain sub-optimal for investigation of viruses. This project uses high-throughput platforms for in vitro investigation of virus entry mechanisms in the context of SARS-CoV-2 and its variants. The outputs of these experiments can, in turn, be used to train and optimise ML language models to build better investigative tools and reveal fundamental biology. The ultimate goal of this work would be to build systems that can determine viral fitness from protein sequence alone and predict the future evolution of viral variants.
Training: This is a truly multidisciplinary project and the student will gain a valuable foundational training in basic virology and computational biology, making them highly competitive in a future career. Specifically, the project will employ innovative pseudovirus systems that allow high-throughput characterisation of mutations in the spike glycoprotein. These experiments will produce thousands of viral sequences with matched phenotypic data (i.e. a measure of their ability to perform virus entry). These datasets can be analysed with pre-existing ML approaches (for instance to explore protein structure), and can also be used to train new optimised ML models with increased predictive power. The application of ML to molecular virology is in its infancy and we expect that significant impact can be achieved in a short period.
Environment: The Centre for Virus Research is a highly-collaborative institute with a critical mass of investigators with different perspectives. The Grove Lab is broadly interested in viral entry mechanisms with a particular focus on membrane fusion glycoproteins. These proteins are essential for the invasion of virus particles in to host cells and are targeted by host antibody-responses; therefore, investigations in this space are directly relevant to vaccine development. The Grove Lab uses a range of approaches from basic virology, structural biology, advanced imaging and computational biology. We are dedicated to the support and training of junior investigators towards their career objectives.
Rotation: in an initial rotation project the student will gain experience in basic lab and computational approaches. Making use of pre-existing data, we will perform pilot ML analyses and test predictions in the lab. We can further tailor the rotation project to fulfil the interests of the student. Please reach out and arrange a meeting if our work is of interest.