Skip to main content
eScholarship
Open Access Publications from the University of California

UC San Diego

UC San Diego Electronic Theses and Dissertations bannerUC San Diego

Multi-Method Analysis of Cellular Composition and Cell–Cell Communication with MRI-Coupled Single-Cell Analysis

Abstract

This thesis evaluates computational approaches for characterizing cell composition alterations and applies an integrated computational framework to investigate cell–cell communication (CCC) in brain cortical organoids (BCOs) derived from autism spectrum disorder (ASD) donors, linking CCC changes to donor-specific brain structure measured by magnetic resonance imaging (MRI). To assess cell type abundance-related bias that can confound CCC inference, a dedicated cell composition analysis framework was adapted, in which single-nucleus RNA sequencing (snRNA-seq) data of patient-derived BCOs were analyzed using three complementary composition-scoring methods: raw cell-type frequency, MELD (Manifold Enhancement of Latent Dimensions), and scCODA (Single-cell Compositional Data Analysis).First, the three approaches were compared to characterize cell type composition differences between ASD and typically developing (TD) BCOs (N = 10 donors), testing whether compositional shifts were recurring across methods with different sensitivities to abundance-related bias. Truncated radial glia (tRG) emerged as a population of interest across the three approaches, showing lower relative abundance in ASD, differences in MELD condition-likelihood scores, and elevated posterior inclusion probabilities in scCODA. Next, MRI-informed CCC analysis was performed in ASD BCOs using LIANA (Ligand-receptor Interaction ANAlysis) and Tensor-cell2cell to identify signaling programs associated with donor brain volume across two organoid ages. Recurrent programs included NLGN-NRXN interactions, the MDK-PTPRZ1 trophic axis, axon guidance, extracellular matrix and integrin signaling, and WNT-associated pathways. In an exploratory cross-modal analysis, larger BCO size was inversely associated with default mode network (DMN) volume, suggesting a relationship between prenatal organoid growth dynamics and postnatal cortical network morphology. Although composition scores were retained as an exploratory analysis rather than integrated into the final CCC inference pipeline of Tensor-cell2cell, these findings provide a proof of concept for multi-method assessment of cell composition bias prior to CCC analysis and suggest that altered progenitor composition and intercellular signaling may contribute to organoid overgrowth and neuroanatomical variation in ASD.