A Community Effort to Translate Protein Data to Knowledge: An Integrated Platform

描述（由申请人提供）：BD2K计划的启动证明了NIH和我们社区的远见卓识。显然，生物医学的未来取决于我们将大数据转化为可理解的科学事实的集体能力。与BD2K的目标一致，我们的目标是彻底改变我们应对从无序数据中识别意义的普遍挑战的方式。利用我们的研究人员在计算生物学和心血管医学方面的互补优势，我们将展示基于心血管研究重点的尖端创新融合，包括：(i)云上数据处理，（ii）人群外包和文本挖掘数据注释，（iii）蛋白质时空动力学，（iv）多组学集成，(v)多尺度临床数据建模。根据我们在创建和完善生物信息学工具方面的十年经验，我们建议通过一个新的查询系统和云基础设施来访问多个组学存储库，并通过计算支持的众包计划来挖掘生物医学文献，将已建立的大数据资源合并为一个可用于数据注释和协作研究的通用模型。我们建议通过机器学习方法构建分子数据并定义与药物和疾病的关系，以及通过云上集成和多维分子数据路径可视化的新算法，将不同的数据类型交织在一起，以揭示从多个尺度上的分子实体合并而成的生物网络。此外，我们建议创新先进的建模工具来解决蛋白质动力学和时空分子机制，通过蛋白质特性的机制建模和3D蛋白质表达图，并通过贝叶斯算法关联患者表型，健康史和多尺度分子谱。我们的工具对更广泛的研究人群的实用性和可定制性是通过三个典型的工作流程来清楚地展示的，这些工作流程可以对大量的基因、转录本、蛋白质或代谢物进行注释；随着时间的推移获得的复杂蛋白质数据集的强大分析；以及对各种分子、文本和文献数据的无缝调查。这些工作流程将使用来自Jackson心脏研究和健康老年人长寿（Wellderly）两个重要临床队列的数据进行严格验证。与此同时，将实施一项多方面的战略，以教育和培训生物医学研究人员，并让公众参与推动整个BD2K计划。

DESCRIPTION (provided by applicant): The inception of the BD2K Initiative is a testament to the foresight of NIH and our community. Clearly, the future of biomedicine rests on our collective ability to transform Big Data into intelligible scientific facts. In line with the BD2K objectives,our goal is to revolutionize how we address the universal challenge to discern meaning from unruly data. Capitalizing on our investigators' complementary strengths in computational biology and cardiovascular medicine, we will present a fusion of cutting-edge innovations that are grounded in a cardiovascular research focus, encompassing: (i) on-the-cloud data processing, (ii) crowd sourcing and text-mining data annotation, (iii) protein spatiotemporal dynamics, (iv) multi-omic integration, and (v) multiscale clinical data modeling. Drawing from our decade of experience in creating and refining bioinformatics tools, we propose to amalgamate established Big Data resources into a generalizable model for data annotation and collaborative research, through a new query system and cloud infrastructure for accessing multiple omics repositories, and through computational-supported crowdsourcing initiatives for mining the biomedical literature. We propose to interweave diverse data types for revealing biological networks that coalesce from molecular entities at multiple scales, through machine learning methods for structuring molecular data and defining relationships with drugs and diseases, and through novel algorithms for on-the-cloud integration and pathway visualization of multi-dimensional molecular data. Moreover, we propose to innovate advanced modeling tools to resolve protein dynamics and spatiotemporal molecular mechanisms, through mechanistic modeling of protein properties and 3D protein expression maps, and through Bayesian algorithms that correlate patient phenotypes, health histories, and multi-scale molecular profiles. The utility and customizability o our tools to the broader research population is clearly demonstrated using three archetypical workflows that enable annotations of large lists of genes, transcripts, proteins, or metabolites; powerful analysis of complex protein datasets acquired over time; and seamless aQoregation of diverse molecular, textual and literature data. These workflows will be rigorously validated using data from two significant clinical cohorts, the Jackson Heart Study and the Healthy Elderly Longevity (Wellderly). In parallel, a multifaceted strategy will be implemented to educate and train biomedical investigators, and to engage the public for promoting the overall BD2K initiative. We are convinced that a community-driven BD2K initiative will best realize its scientific potential and transform the research culture in a sustainable manner, exhibiting lasting success beyond the current funding period.