Methodology development and application to analyse spatial transcriptomics data
| dc.contributor.advisor | Huang, Jian | |
| dc.contributor.advisor | Cronin, Michael | |
| dc.contributor.author | Wang, Qingyue | en |
| dc.date.accessioned | 2026-05-20T09:34:32Z | |
| dc.date.available | 2026-05-20T09:34:32Z | |
| dc.date.issued | 2025-12-31 | |
| dc.date.submitted | 2025-12-31 | |
| dc.description.abstract | Spatial transcriptomics is a rapidly developing technology that enables high-throughput profiling of gene expression at defined spatial locations within intact tissues, thereby linking molecular measurements with the histological context. By enabling the study of cells in their native microenvironment, these technologies provide new opportunities to investigate spatial heterogeneity, cell-to-cell interactions, and the organization of complex biological systems. However, analytical challenges such as mixed-cell capture spots, sparse gene expression, and limited integration with proteomic data necessitate the development of dedicated statistical methodologies. The overarching aim of this thesis is both methodological development and application. To address data sparsity and improve cell type identification, this thesis introduces a clustering-based deconvolution framework that explicitly incorporates spatial coherence to inform deconvolution strategies. In addition, the analytical scope of spatial transcriptomics is extended through the development of predictive modeling approaches to infer protein abundance from transcriptomic data, thereby offering a pathway toward spatial multi-omic integration. Large-scale benchmarking and systematic validation across multiple experimental datasets provide a rigorous evaluation of performance, generalizability, and computational trade-offs for a diverse set of statistical and machine learning models. Applications to real biological data illustrate the ability of these methods to delineate spatially organized domains, characterize protein-specific variability in predictability, and generate biologically interpretable hypotheses on tissue organization and cellular behavior. In conclusion, this thesis advances the statistical foundations of spatial transcriptomics by introducing noval methodologies that address core analytical limitations, establishing frameworks for rigorous model evaluation, demonstrating how these approaches can yield biologically meaningful insights. Collectively, these contributions help transform spatial transcriptomics into a more robust and comprehensive platform for integrative and quantitative analysis in biomedical research. | en |
| dc.description.status | Not peer reviewed | en |
| dc.description.version | Accepted Version | en |
| dc.format.mimetype | application/pdf | en |
| dc.identifier.citation | Wang, Q. 2025. Methodology development and application to analyse spatial transcriptomics data. PhD Thesis, University College Cork. | |
| dc.identifier.endpage | 152 | |
| dc.identifier.uri | https://hdl.handle.net/10468/18793 | |
| dc.language.iso | en | |
| dc.publisher | University College Cork | en |
| dc.rights | © 2025, Qingyue Wang. | |
| dc.rights.uri | https://creativecommons.org/licenses/by-nc-nd/4.0/ | |
| dc.subject | Biostatistics | |
| dc.subject | Bioinformatics | |
| dc.subject | Spatial transcriptomics | |
| dc.subject | Cancer | |
| dc.title | Methodology development and application to analyse spatial transcriptomics data | |
| dc.type | Doctoral thesis | en |
| dc.type.qualificationlevel | Doctoral | |
| dc.type.qualificationname | PhD - Doctor of Philosophy |
