Abe Hanna, Lin Phillip, Zhou Dan, Ruderfer Douglas M, Gamazon Eric R
Vanderbilt University, Nashville, TN, USA.
Division of Genetic Medicine, Department of Medicine, Vanderbilt University Medical Center, Nashville, TN, USA.
HGG Adv. 2025 Apr 10;6(2):100397. doi: 10.1016/j.xhgg.2024.100397. Epub 2024 Dec 31.
Single-cell transcriptome data can provide insights into how genetic variation influences biological processes involved in human physiology and disease. However, the identification of gene-level associations in distinct cell types faces several challenges, including the limited reference resources from population-scale studies, data sparsity in single-cell RNA sequencing, and the complex cell state pattern of expression within individual cell types. Here, we develop genetic models of cell-type-specific and cell-state-adjusted gene expression in mid-brain neurons undergoing differentiation from induced pluripotent stem cells. The resulting framework quantifies the dynamics of the genetic regulation of gene expression and estimates its cell-type specificity. As an application, we show that the approach detects known and new genes associated with schizophrenia and enables insights into context-dependent disease mechanisms. We provide a genomic resource from a phenome-wide application of our models to more than 1,500 phenotypes from the UK Biobank. Using longitudinal, genetically determined expression, we implement a predictive causality framework, evaluating the prediction of future values of a target gene expression using prior values of a putative regulatory gene. Collectively, the results of this work demonstrate the insights that can be gained into the molecular underpinnings of disease by quantifying the genetic control of gene expression at single-cell resolution.
单细胞转录组数据能够为基因变异如何影响人类生理和疾病相关的生物学过程提供见解。然而，在不同细胞类型中识别基因水平的关联面临若干挑战，包括来自群体规模研究的参考资源有限、单细胞RNA测序中的数据稀疏性以及单个细胞类型内复杂的细胞状态表达模式。在此，我们构建了从中脑神经元从诱导多能干细胞分化过程中特定细胞类型和细胞状态调整的基因表达的遗传模型。由此产生的框架量化了基因表达遗传调控的动态变化，并估计其细胞类型特异性。作为一项应用，我们展示了该方法能够检测出与精神分裂症相关的已知和新基因，并有助于深入了解依赖于背景的疾病机制。我们提供了一个基因组资源，该资源来自于我们的模型在英国生物银行1500多种表型上的全表型应用。利用纵向的、基因决定的表达，我们实施了一个预测因果关系框架，使用假定调控基因的先前值来评估目标基因表达未来值的预测。这项工作的结果共同表明，通过在单细胞分辨率下量化基因表达的遗传控制，可以深入了解疾病的分子基础。
