Developing an Integrative Spatial Multi-Omics Pipeline to Map Molecular Alterations in Murine Models
Date Issued
May 2026
Author(s)
Jarrahi, Amin
The University of Tennessee
Advisor(s)
Colleen Crouch
Additional Advisor(s)
Hairong Qi
Bryan Good
Sara Hanrahan
Abstract
Understanding spatial heterogeneity in biological tissues is essential for understanding the molecular processes that contribute to complex diseases. Spatial omics technologies such as mass spectrometry imaging (MSI) and spatial transcriptomics make it possible to map lipids, metabolites, and RNA transcripts directly within tissue sections while preserving spatial information. However, the high dimensionality of these datasets and the differences in spatial resolution and sampling geometry between technologies create significant analytical and integration challenges.
This dissertation presents the development of computational frameworks for analyzing and integrating spatial omics data, with a focus on lipidomics and transcriptomics in the context of aging and Alzheimer’s disease (AD). AD was selected as a biological model due to its strong associations with lipid dysregulation, gene expression changes, and spatially localized pathology in the brain.
First, a Python-based analytical method was developed to perform analyte preprocessing and quantitatively assess analyte delocalization in matrix-assisted laser desorption/ionization (MALDI) MSI data. A composite delocalization score was introduced that integrates normalized off-tissue area and mean distance from the tissue boundary, providing a robust measure of spatial spreading beyond tissue regions. This metric enables systematic evaluation of analyte reliability and supports quality control in MSI experiments.
Second, a computational framework was created to enable large-scale exploration of MSI datasets. The pipeline integrates preprocessing, dimensionality reduction, clustering, and statistical analysis to identify molecular features exhibiting biologically meaningful spatial patterns. Application to MALDI-MSI lipidomics data from mouse brain tissues representing young, aged, and AD models revealed region-specific lipid variations and candidate analytes associated with aging and disease.
Third, a Python-based workflow for spatial transcriptomic analysis using the 10x Genomics Visium platform was developed. The pipeline successfully performed preprocessing, clustering, and differential expression analysis while preserving spatial context, identifying genes associated with both AD and aging.
Finally, a multi-objective scoring (MOS) framework was developed to integrate spatial transcriptomics and MSI data without requiring exact spatial co-registration. The framework combines multiple spatial descriptors to identify lipid features with transcriptomic patterns. Together, these tools provide a scalable framework for spatial multi-omics analysis and hypothesis generation in complex biological systems.
Degree
Doctor of Philosophy
Major
Biomedical Engineering
File(s)![Thumbnail Image]()
Name
JARRAHI-PRIMARY-2026.pdf
Size
96.15 MB
Format
Adobe PDF
Checksum (MD5)
e62ab505849158da0ab51ee6559fd212