Multiscale transcriptional architecture of the human brain

在美国，近一半的人会在一生中的某个时候达到精神障碍的诊断标准。 尽管大多数精神障碍是中度到高度可遗传的，但它们潜在的遗传结构是 很复杂。这种复杂性阻碍了为患者确定生物标记物和开发诊断测试和治疗策略的努力。尽管有这种复杂性，但我们有理由预计，基因变异 易患精神障碍的人会改变大脑中的基因表达。许多研究已经尝试过 为了在死后大脑样本中确定精神障碍的转录表型，但使用批量组织导致了样本和数据集之间的细胞组成不同，掩盖了转录表型及其细胞来源。单细胞和单核转录的研究进展 分析为这个问题提供了一种新的方法，但这些技术也有技术和统计上的限制。因此，在我们对精神障碍的风险因素如何扰乱的理解上仍然存在严重的差距 人类大脑中的基因表达。识别此类扰动的一个主要障碍是缺少 用于预测人脑中基因表达的分析框架，而不考虑采样策略。这个 这一应用的目的是检验神经生物转录本的协方差结构提供了这样一个框架的假设。在目标1中，我们将评估来自人脑样本的大量和SC/SN基因表达数据中细胞类型特征和转录协变的一致性。在目标2中，我们将 阐明最佳实验和分析参数，以确定可重现的转录签名 人脑样本中罕见的细胞类型/状态。在目标3中，我们将利用批量和SC/SN基因表达数据中可重现的转录共变来鉴定精神障碍的细胞和分子表型 以及相应的遗传变异。总的来说，拟议的研究将产生积极的影响，通过促进 通过荟萃分析和预测建模在神经生物学研究中的严密性和重复性，同时提出识别精神障碍的细胞和分子表型的新策略 可以很容易地应用于其他分子物种和病理条件。

In America, nearly half of individuals will meet diagnostic criteria for a mental disorder at some point in their life. Although most mental disorders are moderately to highly heritable, their underlying genetic architecture is complex. This complexity has hindered efforts to identify biomarkers and develop diagnostic tests and therapeutic strategies for patients. Notwithstanding this complexity, it is reasonable to expect that genetic variants that predispose individuals to mental disorders will alter gene expression in the brain. Many studies have tried to identify transcriptional phenotypes of mental disorders in postmortem brain samples, but the use of bulk tissue has resulted in variable cellular composition across samples and datasets, obscuring transcriptional phenotypes and their cellular origins. Recent advances in single-cell (SC) and single-nucleus (SN) transcriptional profiling offer a new approach to this problem, but these techniques also have technical and statistical limitations. As such, there remain critical gaps in our understanding of how risk factors for mental disorders perturb gene expression in the human brain. A major impediment to identifying such perturbations is the absence of an analytical framework for predicting gene expression in the human brain regardless of sampling strategy. The purpose of this application is to test the hypothesis that the covariance structure of neurobiological transcriptomes provides such a framework. In Aim 1, we will assess the concordance of cell-type signatures and transcriptional covariation in bulk and SC/SN gene expression data from human brain samples. In Aim 2, we will clarify optimal experimental and analytical parameters for identifying reproducible transcriptional signatures of rare cell types/states in human brain samples. And in Aim 3, we will exploit reproducible transcriptional covariation in bulk and SC/SN gene expression data to identify cellular and molecular phenotypes of mental disorders and corresponding genetic variants. Collectively, the proposed studies will have a positive impact by promoting rigor and reproducibility in neurobiological research through meta-analysis and predictive modeling, while simultaneously advancing new strategies for identifying cellular and molecular phenotypes of mental disorders that can be readily applied to other molecular species and pathological conditions.