Interpretable approaches for cellular organisation and molecular profiles using (spatial) omics data in health and disease

Ali, Mayar (2025): Interpretable approaches for cellular organisation and molecular profiles using (spatial) omics data in health and disease. Dissertation, LMU München: Graduate School of Systemic Neurosciences (GSN)

Abstract

Deciphering cellular organisation and how molecular expression patterns vary across tissues is fundamental to gaining insights into biological processes and disease mechanisms. Advances in omics technologies, particularly spatial omics, have revolutionised our ability to investigate these patterns at unprecedented resolution and dimensionality. While various statistical and machine-learning approaches have emerged to analyse these complex data sets, critical gaps remain in our understanding of the differential cellular organisation and tissue characteristics across different conditions. Furthermore, the field typically lacks robust tools for a three-dimensional, holistic view of tissues, which is crucial for comprehending disease dynamics. In this thesis, I present a suite of mathematical approaches to analyse diverse omics data, focusing on spatial omics, with various resolutions, throughputs, and dimensionalities to reveal crucial insights into differential changes in molecular expression and cellular organisation across conditions. First, as part of a large multimodal comprehensive study, I expanded on our understanding of cell heterogeneity, the distinctive proteome signatures in the skull bone marrow, and its role in immunological responses to neurological disorders. Second, increasing the complexity and advancing into the spatial omics domain, I advanced the current approaches to quantitatively analyse the cellular organisation across multiple scales, from individual cell types and tissue niches to whole tissue revealing condition-specific tissue changes. Third, I proposed graph models and comprehensive multimodal ablation studies to understand the tissue traits contributing to patient outcomes and associate them with tissue architecture motifs to enhance our understanding of disease progression. Fourth, moving into three-dimensional space, I used our new technology, DISCO-MS, to explore the proteome changes in amyloid-beta plaques in Alzheimer’s disease to capture very early (6 weeks) and late-stage dynamics and region-specific variations, providing a holistic view of the plaques’ microenvironment. Collectively, these approaches represent a comprehensive effort and advancement in our ability to study the differential changes in molecular expression, cellular organisation, and tissue traits across different physiological and pathological conditions while further extending into three-dimensional volumes for a more holistic understanding of biological systems.

• Dokumententyp: Dissertationen (Dissertation, LMU München)
• Keywords: Spatial omics, proteomics, machine learning, graph neural networks, tissue clearing, statistical modelling
• Themengebiete: 500 Naturwissenschaften und Mathematik ; 500 Naturwissenschaften und Mathematik > 570 Biowissenschaften, Biologie
• Fakultäten: Graduate School of Systemic Neurosciences (GSN)
• Sprache der Hochschulschrift: Englisch
• Datum der mündlichen Prüfung: 2. Juni 2025
• 1. Bericht­erstatter:in: Ertürk, Ali