Background: Human genetic studies have successfully identified thousands of genomic regions associated with common diseases, but a major challenge is to pinpoint the specific causal gene within each region and the exact cellular environment where it acts. This project will develop a novel Artificial Intelligence system to solve both problems simultaneously by integrating human genetics with revolutionary new high-resolution maps of gene activity in human tissues (spatial transcriptomics).
Novelty and Importance: The key innovation is to reframe this biological puzzle as a search problem solved using a Genetic Algorithm, a type of AI inspired by Darwinian evolution. Our algorithm will `evolve` a population of candidate solutions, where each solution is a set of potential causal genes. The `fittest` solutions are those where all the genes are found to be highly active in the same, very specific cellular neighbourhood. This novel AI-driven search strategy moves beyond simple statistics to a model-based inference of causality.
Aims and Objectives: The primary aim is to engineer a robust and interpretable AI framework for causal discovery in biology. The student will first develop the methods to integrate genetic and spatial transcriptomics data. They will then design and build the core Genetic Algorithm. Finally, they will apply and validate the system on a complex human disease, with the goal of delivering a ranked list of high-confidence causal genes and the specific cellular niches where they function, providing powerful new insights for therapeutic development.