Spatial genomics technologies to study cancer and genetic diseases in tissue contexts
Aim
The general aim of the proposed project is to develop an innovative spatial technology with a complete pipeline from wet-lab experiments to dry-lab analysis to study tissue biology of cancer and genetic diseases. Most genomics technologies require disassociation of cells from original tissues, thereby discarding the physiological spatial information. The spatial technology allows to quantitatively assess tissue heterogeneity in cancer and genetic underpinnings of disease at tissue and single-cell level without dissociating the tissue. With both spatial image and genomics information, the technology allows to develop innovative software tools for combining image and transcriptional genomics data that will eventually be useful for computer-aided tissue diagnosis.
Brief project outline
Combining gene expression data with the location and physiological context of the cells in a tissue, we will determine cell types, groups of cells spatially clustered into microenvironments, and molecular interactions between cell types. To integrate genomics (microscopic) and imaging (macroscopic) data for systematically assessing diseased tissues, we will develop an innovative analysis approach using a convolutional neural network framework, which is especially suitable for spatial data, and fast parallel computation to train multiple layers of features in image. Through this project, we aim to evaluate the use of spatial transcriptomics together with machine learning to better define the molecular nature of the tissue heterogeneity in cancer and genetic diseases. We expect that spatial transcriptomics will bring the world-class technology that can serve the broad genomics research projects at UQ and beyond.
Genomics-based innovative aspect of proposal
We aim to develop spatial transcriptomics (ST) to become a major new capability that will help UQ to establish a pioneering nationally and competitive internationally position in genomics technology. Current genomics assays are limited by: 1) the need to dissociate cells from native tissues, thus discarding the in vivo spatial context, 2) the lack of resolution to measure at the single-cell level, and 3) the number of genes that can be measured in situ. In a world-first, we will combine the state-of-the-art ST together with the rapid advance in artificial intelligence analytics. For complex, multimodal data analysis, we develop an innovative approach to integrate genomics (microscopic) and imaging (macroscopic). The approach therefore integrates methods from a variety of experimental and data analytics fields using cutting-edge genomics technology and advanced computational analysis that not only allow UQ to stay at the spearhead of dry and wet genomics research.
Broad applicability of the technique
We expect that spatial transcriptomics will bring the world-class technology that can serve the broad genomics research projects at UQ and beyond. We expect that the technology can be transferred to IMB sequencing facility as a complete pipeline from sequencing to data analysis, thereby making it widely available and implementable to UQ researchers. Users only need to submit fresh-frozen tissues or OCT embedded tissues. The data can be readily incorporated with other genomics data.
Developments arising from this project
GIH development: RNAscope® HiPlex assay using an automated slide scanner
Bioinformatic tool: stLearn - to comprehensively analyse Spatial Transcriptomics (ST) data to investigate complex biological processes within an undissociated tissue.
Publications arising from this project
A robust experimental and computational analysis framework at multiple resolutions, modalities and coverages (https://doi.org/10.3389/fimmu.2022.911873)
Spatially resolved transcriptomes of healthy mammalian kidneys illustrate the molecular complexity and interactions of functional nephron segments (https://doi.org/10.3389/fmed.2022.873923)
SpaCell: integrating tissue morphology and spatial gene expression to predict disease cells (https://doi.org/10.1093/bioinformatics/btz914)
Pre-print: stLearn: integrating spatial location, tissue morphology and gene expression to find cell types, cell-cell interactions and spatial trajectories within undissociated tissues (https://doi.org/10.1101/2020.05.31.125658)
