Grant: Discovery Grants Program - Individual

Spatial transcriptomics (ST) is a new technology that enables researchers to take pictures of tissues and simultaneously measure gene activity at high resolution across the tissue. This technology has been used to map the location of different types of neurons across the layers of the brain, and all the cells of mouse embryos. Combining spatial information with the number of RNA copies present for all genes in the genome will enable a greater understanding of tissue development and tissue structure than was previously possible, but only if appropriate statistical tools are available to analyze this data. My research program will develop a suite of data analysis tools to enable the efficient use of spatial transcriptomic data to answer important biological questions. A key question that is now answerable with ST is whether two cell-types or genes are present in the same location in the tissue, i.e. spatial colocalization. For instance, whether two types of immune cells migrate together into tissues during inflammation, or whether a specific signaling pathway in one cell may induce its neighbour to become a stem cell. However, this is challenging because colocalization may be specific to certain parts of a tissue. This project will adapt spatial correlation statistics and regression models to be applicable to ST data to robustly determine where genes colocalize. Another key application is determining how tissue structure changes across an organism's lifespan or in response to a biological process such as symbiosis or stress. To evaluate these shifts researchers need a method to identify regions that are equivalent across different time points or conditions, and regions that are unique. We will develop a method that uses mutual nearest-neighbour networks that incorporate spatial location to identify matching tissue regions. To make full use of all aspects of ST data we will use machine learning to reduce complex tissues images into a small number of key features that can be more easily combined with measurements of gene activity. These features will be used to build an image-based network that can be seamlessly combined with gene-activity networks in an easily interpretable and flexible manner.  These tools will accelerate all kinds of biological research both academic and industrial that will benefit from the use of spatial transcriptomics technologies. From understanding how microbes interact with plant roots to accelerate plant growth, to understanding how the structure of plant leaves change in response to stresses, to unraveling the evolution of metazoan complexity, to a better understanding of how the human body repairs and maintains itself. By focusing on unsupervised approaches to gene-expression and image analysis, I will enable the application of these methods to a broader array of species and tissues that are of interest to many industries, not simply humans and model organisms.