Preview

PDF, English Download (15MB) | Terms of use

Abstract

The impact of genetic variation on molecular traits such as gene expression is known to vary substantially across molecular contexts, such as cell types and states. Conventional approaches for molecular phenotyping rely on bulk sequencing data, representing aggregate measurements across thousands or millions of cells from sorted cell populations or whole tissue sections. As a result, these methods often lack resolution to capture fine-grained cellular heterogeneity, impacting discovery and explanatory power for mapping genetic associations.

More recently, single-cell sequencing technologies have fundamentally changed our ability to study cellular identity. By isolating and analyzing individual cells, these methods have revealed remarkable diversity even within seemingly homogeneous cell populations. Integrating single-cell measurements with genotyping data opens up new opportunities for linking genetic variation to molecular processes in a context-specific manner, aiding disease diagnosis and therapeutic design. However, existing analysis strategies typically aggregate measurements in discrete cell clusters, potentially missing subtle allelic regulation and interaction effects with continuous biological processes such cell differentiation or development. This thesis proposes three new computational methods to model genetic effects at the level of individual cells, to leverage the full potential of single-cell measurements.