On the Form and Function of Single Cells

Abstract

Engineering biological systems requires understanding form-function relationships at the scale of the single cell. Recent developments in genomics facilitate the characterization of single-cells at multiple resolutions, but generating insight from these data remains challenging. We need frameworks to describe, integrate, and contextualize genomic features across scales. This dissertation investigates form-function relationships in single-cells: from the architecture of the genome within nucleus to the interactions between cells within tissues. We introduce a computational method to identify critical higher-order chromatin structures from Pore-C and gene expression data and leverage these structures to improve the signal-to-noise ratio in single-cell Pore-C data. We utilize long-read single-cell transcriptomics data to characterize direct reprogramming of fibroblasts into a hematopoietic stem cell-like phenotype and propose a computational approach to assess the phenotypic similarity of reprogrammed cells to their initial and target states. We present a framework for analyzing spatial transcriptomics data over time and apply this framework to study immune cell dysregulation in adipose tissue during obesity progression. At a broader scale, we present a programmable platform for wound healing assays, which enables the study of single-cell migration, proliferation, and cell-cycle dynamics using live-cell fluorescence microscopy data. Finally, we propose algorithms inspired by affinity maturation in the adaptive immune system, which serves as a foundation for immune-inspired machine learning.