INSPiRe · Graph Deep Learning for detection of INtracellular and SPatial interactions Revealing niches

Tissues are multicellular systems organized into niches, highly specialized communities of spatially co-localized cells (spatial neighborhood) that interact with one another (intercellular signaling), shaping their coordinated functions (intracellular states). Advances in single-cell sequencing and spatially resolved technologies now enable the reconstruction of whole-tissue spatial atlases spanning millions of cells and thousands of molecular features (e.g., transcriptomics, epigenomics). However, identifying functional niches from these data remains challenging due to the complexity and multidimensionality of biological processes, as well as the scale of modern datasets. INSPiRe project proposes a novel graph deep learning–based computational framework that integrates single-cell multi-omics with spatial transcriptomics to systematically identify and characterize cellular niches. Unlike existing approaches, our method jointly encodes three complementary dimensions of tissue organization: (i) cellular neighborhood and proximity, (ii) intercellular communication via ligand–receptor interactions, and (iii) intracellular, pathway-specific signaling activity, a combination not addressed by current tools. Leveraging adversarial learning and efficient graph representations, the framework generates interpretable and scalable cell embeddings from large multi-tissue datasets, quantitatively capturing niche structure and providing a functional understanding of how cellular niches emerge within tissue microenvironments. This is a timely and innovative project that will advance our understanding of tissue biology and disease mechanisms, with applications across healthy, developmental, and diseased tissues. Moreover, the project will provide the newly-formed researcher hands-on experience in advanced computational methods, gaining in-depth knowledge of modeling complex systems, while also developing project management and leadership skills, through a structured training plan.