Integrating Transcriptomic and Neuroimaging Data to Uncover Genetic Underpinnings of Human Brain Networks

Abstract

Brain networks facilitate human sensation, cognition, and behavior. The organization of these networks varies substantially between individuals. Much of this variation is strongly heritable, suggesting that it is shaped by genetically regulated gene expression. Advances in neuroimaging and genomics have allowed us to study brain organization and gene expression at a large scale. However, the specific mechanisms linking the expression of groups of genes to the complex organization of brain networks remain largely unmapped. Specifically, there are still unanswered questions about which genes influence human brain connectivity and how they regulate the complex interactions within and between brain regions. The work in this thesis aims to bridge this gap by integrating transcriptomic and neuroimaging data to uncover these relationships. To achieve this, we leveraged large-scale multimodal data, integrated imputed brain gene expression with MRI-based measures of regional structure and network connectivity and identified patterns of gene expression that are statistically tied to aspects of brain organization. In this way, our work delineates distinct biological pathways that drive neuroimaging network phenotypes. Our specific approach comprised several steps. First, we quantified regional structural and functional phenotypes using UK-Biobank structural and functional MRI, as well as network-level phenotypes from diffusion MRI. Next, we developed and evaluated a non-negative regularized integrative framework based on Multiview Canonical Correlation Analysis with recursive feature pruning to identify stable cross-modal patterns that jointly relate gene expression, regional morphology, and network organization in a shared latent space. Last, we interpreted these patterns biologically and clinically by examining overlap with the findings of previous genome-wide association studies, Vanderbilt’s BioVU resource, and by mapping selected genes onto comprehensive human brain cell-type atlases to identify the cellular pathways underlying the large-scale systems we uncover. Together, our findings bridge studies of the genome and the connectome by identifying key mechanisms underlying individual variation in these substrates in brain health and disease.