1/2-Cis & Trans-Data Integration to Find Mechanisms Causing Psychiatric Disorder

描述（由申请人提供）：与精神疾病的生物学原因相关的数据量呈指数级增长。整合所有这些研究的结果可以通过增加发现生物标志物的统计能力，减少由于平台特定技术错误而导致的错误发现，以及在多条证据指向相同生物因素时增加对结果的信心，从而推进心理健康研究。由于数据整合可能涉及异构数据集和生物学关系，因此它也具有相当大的潜力，可以通过阐明生物因素相互作用的更广泛背景来提高我们对疾病机制的理解。这项工作建议建立在Van den Oord （VCU）和Sullivan （UNC）实验室的长期合作和互补专业知识的基础上。结果是一个高度连贯、严谨和创新的数据整合策略，旨在提高对精神分裂症、双相情感障碍、重度抑郁症和自闭症的独特和共同疾病机制的理解。具体来说：(1)我们将使用的数据来源是深入的、全面的，并且是为精神病学量身定制的；(2)我们开发了MIND包（基于数学的异构数据集成），它基于严格的数学框架，允许以有意义的和统计上最优的方式进行数据集成。MIND没有完全依赖于模拟，而是通过一项大型独立复制研究进行了实证测试，结果表明，它识别出了生物标志物，否则需要更多的样本或遗传标记；(3)由于我们开发了可以在所有可用数据集中对几乎任何种类的生物学关系进行独立测试的方法，因此可以在非常大的样本中研究疾病机制；(4)我们将通过SLEP（沙利文实验室证据项目）为高级用户提供所有结果和软件，并以用户友好的方式为最终用户提供。拟议项目的成功完成将(a)使最终用户能够随时获得用于数据整合的复杂工具，(b)使高级用户能够根据自己的选择调整使用这些资源，以及(c)为更好地了解精神障碍的疾病机制作出重要的、高影响力的贡献。

DESCRIPTION (provided by applicant): The amount of data related to the biological causes of psychiatric conditions has grown exponentially. Integrating results from all these studies can advance mental health research by increasing statistical power to find biomarkers, reduce false discoveries due to platform-specific technical errors, and increase confidence in findings when multiple lines of evidence point to the same biological factors. Because data integration can involve heterogeneous datasets and biological relations, it also has a considerable potential to improve our understanding of disease mechanisms by elucidating the broader context in which biological factors co-act. This work proposed builds on the long collaboration and complementary expertise of the Van den Oord (VCU) and Sullivan (UNC) labs. The result is a highly coherent, rigorous, and innovative data integration strategy aimed at improving understanding of unique and shared disease mechanisms underlying schizophrenia, bipolar disorder, major depressive disorder, and autism. Specifically: (1) the data sources we will use are deep, comprehensive, and tailored to psychiatry; (2) we developed the MIND package (Mathematically-based Integration of heterogeNeous Data) that is based on a rigorous mathematical framework that allows data integration in a meaningful and statistically optimal fashion. Rather than relying exclusively on simulations, MIND was tested empirically using a large independent replication study showing that it identified biomarkers that would otherwise require far more samples or genetic markers; (3) as we developed methods that can perform independent tests of virtually any kind of biological relationship in all available datasets, diseae mechanisms can be studied in very large samples; and (4) we will make all results and software available via SLEP (Sullivan Lab Evidence Project) for power users and in an user- friendly implementation for end users. Successful completion of the proposed project will (a) allow end users ready access to sophisticated tools for data integration, (b) allow power users to adapt use these resources as they choose, and (c) make an important, high-impact contribution to better understand disease mechanisms underlying psychiatric disorders.