Computational methods for sub-cellular spatial transcriptomics with high gene coverage

Spatial transcriptomics technologies are revolutionizing our understanding of how cells are organized, communicate, synergize to function, and drive tissue phenotypes. Although previously limited by resolution-gene coverage tradeoffs, emerging spatial technologies are enabling genome-wide expression profiling at sub-cellular resolution. With technical platform limitations resolving, new computational methods are urgently needed to realize the full potential provided by high resolution and molecular breadth. Although specialized computational approaches have been developed for previous spatial technologies, the high- dimensionality and sparsity of new types of spatial data introduces significant obstacles to modeling and interpretation. Thus, we here propose to develop advanced machine learning methods and bioinformatics software to discern cell types, discover cell states, and deduce cell-cell interactions from high-resolution, high- dimension spatial transcriptomics data. First, we will develop a series of machine learning methods for supervised cell type mapping by integrating reference single-cell transcriptomics. We will comprehensively evaluate the performance of methods using single-cell RNA-based simulations and validate annotations from experts in tissue histology. Second, we will develop machine learning methods for unsupervised cell state discovery to delineate the spatial organization and potential regulators of cell states beyond cell types. We will benchmark performance and efficiency against other clustering-based tools for spatial transcriptomics data. Finally, we will develop analytical approaches that correct diffusion noise to identify direct cell-cell interactions in high resolution and interaction-associated genes. By developing and implementing these approaches in diverse tissue contexts, such as tumor microenvironments and inflammation, we will provide a robust and generalizable framework for understanding cellular organization, genomic function, and intercellular communications in tissues. 1 Project Number: 1R35GM162431-01 | Fiscal Year: 2026 | NIH Institute/Center: National Institute of General Medical Sciences (NIGMS) | Principal Investigator: Yi Zhang | Institution: DUKE UNIVERSITY, DURHAM, NC | Award Amount: $444,125 | Activity Code: R35 | Study Section: Special Emphasis Panel[ZRG1 MGG-D (55)] View on NIH RePORTER: https://reporter.nih.gov/project-details/11269831