Integration of brain imaging and multi-omics data for improved diagnosis and prediction of mental disorders

标题：整合脑成像和多组学数据以改进诊断和预测 精神障碍的 项目概要 该项目的首要目标是将多尺度组学和脑成像纳入临床研究 对生物学定义的精神疾病进行分类学，并揭示其特定的遗传 架构。这将改变当前精神疾病诊断和预后的实践，从而导致 精准精神病学。我们最近的工作证明了结合多尺度脑成像的价值，例如 作为功能磁共振成像，具有基因组学、网络和生物学知识，可检测风险基因和生物标志物。我们也 开发了一套用于联合功能磁共振成像和基因组学（例如，SNP）分析的综合方法。尽管我们最初 成功后，仍然存在以下重大挑战：1）如何揭示以前隐藏的（例如非线性） 多种数据类型之间的关系，用于检测组学内部和跨组学之间的相互作用网络， 揭示精神疾病的特定遗传结构； 2）如何整合表型相关 多组学分析重新定义和区分临床重叠的多种精神疾病 精神分裂症 (SZ)、双相情感障碍 (BI) 和单相抑郁症 (UD) 等症状； 3）如何链接多 组学和脑成像数据以及表型和认知测量，用于预测临床 结果或疾病状态（预测组）； 4) 如何用新收集的生物标志物验证检测到的生物标志物 通过纳入额外的大脑和组学数据（例如 DTI、甲基化）进行大型队列研究。 在本提案中，我们将解决神经科学和临床精神病学中的上述剩余挑战。 我们将继续提高由图像分析师组成的多学科研究团队的生产力 生物信息学家（Wang）、MRI 成像科学家（Calhoun）、遗传学家和生物统计学家（Deng）以及 精神科医生（皮尔森）。为了利用我们过去的成功，我们建议实现以下具体目标：1) 检测多模式脑成像和基因组学之间复杂的疾病特异性非线性关系 数据并进一步识别组学水平内部和跨组学水平的相互作用网络； 2）合并表型- 将特定的网络和结构信息纳入我们的集成模型中，以检测生物标志物并进一步 在大型数据集上验证它们，以对多种精神障碍及其基因构成进行分类； 3） 将多组学和大脑成像联系起来，包括它们与行为和认知测量的相互作用， 用于预测精神疾病（预测组）； 4）传播综合多组学成像 为神经影像研究社区提供开源软件中的非线性分析分析工具。 这些方法将更好地区分具有重叠症状和症状的精神障碍。 更准确地预测临床结果。这将为精神科的个性化护理奠定基础 通过针对特定的基因组成并评估药物治疗的效果来识别疾病。由 通过向研究界传播软件，该项目将产生广泛且持续的影响。

Title: Integration of brain imaging and multi-omics data for improved diagnosis and prediction of mental disorders Project Summary An overarching goal of this project is to incorporate multiscale omics and brain imaging into clinical studies towards a nosology of psychiatric disorders that are biologically defined, and to uncover their specific genetic architectures. This will transform the current practice of mental disease diagnosis and prognosis, leading to precision psychiatry. Our recent work has demonstrated the value of combining multiscale brain imaging such as fMRI with knowledge of genomics, networks, and biology to detect risk genes and biomarkers. We also developed a set of integrative approaches for joint fMRI and genomics (e.g., SNPs) analysis. Despite our initial success, the following significant challenges remain: 1) how to uncover previously hidden (e.g., nonlinear) relationships among multiple data types for the detection of interaction networks both within and across omics, revealing the specific genetic architecture of psychiatric disorders; 2) how to incorporate phenotype-relevant multi-omics profiling to redefine and differentiate multiple psychiatric disorders with clinically overlapping symptoms such as schizophrenia (SZ), bipolar disorder (BI), and unipolar depression (UD); 3) how to link multi- omics and brain imaging data with phenotypical and cognitive measurements for the prediction of clinical outcomes or disease states (predictome); and 4) how to validate the detected biomarkers with newly collected large cohort studies by incorporating additional brain and omics data (e.g., DTI, methylations). In this proposal we will address the above remaining challenges in neurosciences and clinical psychiatry. We will continue to build on the productivity of our multidisciplinary research team comprising an image analyst and bioinformatician (Wang), an MRI imaging scientist (Calhoun), a geneticist and biostatistician (Deng), and a psychiatrist (Pearlson). To leverage our past success, we propose to accomplish the following specific aims: 1) to detect complex disease-specific non-linear relationships between multi-modal brain imaging and genomics data and further identify interaction networks both within and across omics levels; 2) to incorporate phenotype- specific network and structure information into our integration models for the detection of biomarkers and further validate them on large datasets for the classification of multiple mental disorders and their genetic make-ups; 3) to link multi-omics and brain imaging, including their interactions with behavioral and cognitive measurements, for the prediction of psychiatric disorders (predictome); and 4) to disseminate integrative multi-omics imaging analysis tools featuring non-linear analysis in open source software to the neuroimaging research community. These approaches will lead to better differentiation of mental disorders with overlapping symptomatology and more accurate prediction of clinical outcomes. This will lay the groundwork for personalized care of psychiatric disorders by targeting their specific genetic make-ups and evaluating the effects of medical treatments. By disseminating software to the research community, the project will have a broad and sustained impact.