Enhance UCSC Xena: extend interactive visualization to ultra-large-scale multi-omics data and integrate with analysis resources

摘要 癌症基因组学正在发生两个重大的范式转变：单细胞基因组图谱和生长 NCI癌症研究数据共享中心。单细胞基因组学正在改变我们对 复杂的肿瘤种群，揭示了对肿瘤成分、微环境、癌症干细胞的新见解 细胞，以及抗药性。几个大规模的、以单一单元为重点的国家和国际项目 目前正在进行中，包括HCA、Htan和HuBMAP。这些项目产生的数据将影响到几乎 生物学和医学的方方面面。为了让这些项目充分发挥其潜力， 这是必不可少的 使这些资源可访问的数据可视化和分析工具 给一大群生物医学 研究人员。然而，这是具有挑战性的，因为现有的数据可视化和分析工具根本无法扩展 来处理这些大型数据集。第二个范式转变是NCI对癌症研究数据的发展 Commons(CRDC)，这是一个虚拟数据科学基础设施，它将癌症研究数据收集与 分析工具，利用云的动态计算能力。高效、安全地整合 广泛使用的第三方工具和平台，包括UCSC Xena等交互式可视化工具， CRDC是使这一资源真正有用所必需的。随着这两个过渡在 未来几年，它们带来了挑战和机遇。我们建议增强UCSC Xena以支持和 通过四个目标实现这些转变。目标1.我们将把UCSC Xena扩展100倍，以支持 可视化包含100万个以上细胞(更一般地，100万个生物实体)的数据集，而不会造成任何损失 网络浏览器中的数据或交互性。我们将利用计算机工程中的几个新进展来 实现这一性能提升。此外，我们还将开发三种新的可视化技术，使研究人员能够 更好地探索单元格数据。目标2.我们将安全地将UCSC Xena与NCI CRDC的资源进行整合 及其数据分析工具和平台社区。我们的集成将使装货结束分析 结果放到CRDC中的私有Xena Hub中，以便在大型公共数据的上下文中进行可视化，这是一种常规做法。 目的3.我们将通过扩展提供最新的癌症基因组学资源数据的可视化 以及使用关键项目和数据集更新UCSC Xena数据库。我们将与树屋合作 儿童癌症倡议，以建立统一的临床前儿科基因组数据资源并使其 在Xena浏览器上公开提供。这项工作将利用PDX模型和脑肿瘤有机化合物 目前正在由豪斯勒博士的团队开发和分析。目标4.我们将改进用户工作流程和 通过以用户为中心的设计参与，以及持续的用户教育、支持和推广。

Abstract Two significant paradigm shifts are underway in cancer genomics: single-cell genomic profiling and the growth of the NCI Cancer Research Data Commons. Single-cell genomics is transforming our understanding of complex tumor populations and revealing new insights into tumor composition, microenvironment, cancer stem cells, and drug resistance. Several large-scale, single-cell-focused, national and international projects are currently underway, including HCA, HTAN, and HuBMAP. Data generated by these projects will impact almost every aspect of biology and medicine. For these projects to realize their full potential, it is essential to have data visualization and analysis tools that make these resources accessible to a broad group of biomedical researchers. This is challenging, however, as existing data visualization and analysis tools simply cannot scale to handle these large datasets. The second paradigm shift is NCI’s development of the Cancer Research Data Commons (CRDC), a virtual data science infrastructure that connects cancer research data collections with analytical tools, leveraging the dynamic computing power of the cloud. Efficient and secure incorporation of widely-used 3rd party tools and platforms, including interactive visualization tools such as UCSC Xena, into CRDC is needed to make this resource truly useful. As both of these transitions continue to accelerate in the coming years, they present challenges and opportunities. We propose to enhance UCSC Xena to support and enable these transitions through four aims. Aim 1. We will scale up UCSC Xena by 100x to support the visualization of datasets with greater than 1 million cells (more generally, 1 million bio-entities) without any loss of data or interactivity in the web browser. We will employ several new advances in computer engineering to achieve this performance gain. In addition, we will develop three new visualizations to enable researchers to better explore single-cell data. Aim 2. We will securely integrate UCSC Xena with resources in the NCI CRDC and its community of data analysis tools and platforms. Our integration will make loading ending analysis results into a private Xena Hub in CRDC for visualization in the context of large public data a routine practice. Aim 3. We will provide visualization of the most current cancer genomics resource data through the expansion and update of UCSC Xena database with key projects and datasets. We will collaborate with the Treehouse Childhood Cancer Initiative to build a harmonized preclinical pediatric genomics data resource and make it publicly available on the Xena Browser. This work will leverage PDX models and brain tumor organoids currently being developed and profiled by Dr. Haussler’s group. Aim 4. We will improve user workflows and engagement through User Centered Design, as well as continue user education, support, and outreach.