Data Processing, Analysis and Modeling Unit

项目摘要(数据分析股) 转移性肿瘤是癌症死亡的主要原因，很难治疗。生物学 潜在的细胞状态可塑性和控制适应的不同分子程序 外来微环境将需要对复杂环境有更深入的了解 肿瘤的生长。我们的目标是解决这一重大的知识鸿沟，建立一个时空 三种特别致命的恶性肿瘤--肺癌的转移转移图谱， 胰腺癌和中枢神经系统转移-通过结合单细胞基因组学和多细胞基因组学- 深度注释的患者来源的原发和转移性临床的空间映射 样本。朝着这个目标，我们的第一个目标将是开发实验设计方法，以选择 用于创建这三个图谱的患者、样本和实验参数。这一目标 是基于这样的理论基础，即图谱构建在实验中提出了新的统计挑战 必须开发的设计，以最大限度地利用资源来建设地图集。在我们的 第二个目标是，我们将大规模地建设和实施数据分析单位的基础设施。 这一目标基本原理是质量控制以及可伸缩量化和注释 大规模数据将指导使用地图集进行搜索、比较和解释 样本。最后，我们将开发新的计算方法来进行数据集成和 对时空地图集的解读。在本例中，我们假设来自不同数据 技术和平台通常包括冗余的生物信息，这些信息可以 被提取出来用于制作一份带注释的地图集。我们将说明我们的地图集在使用中的影响 预测早期肺腺癌可能转移至 大脑。最终，构建的地图集将提供关于汇聚事件和瓶颈的洞察 在转移的转变中，暗示了潜在的治疗靶点和机会 干预。

PROJECT SUMMARY (Data Analysis Unit) Metastatic tumors are the leading cause of cancer deaths and are difficult to treat. The biology underlying cell state plasticity and the distinct molecular programs that govern adaptation to foreign microenvironments will require a much deeper understanding of the complex environment of tumor growth. We aim to address this significant knowledge gap by building a spatial-temporal atlas of the metastatic transition of three exceptionally lethal sets of malignancies – lung cancer, pancreatic cancer, and CNS metastases- by combining single-cell genomics with multi- dimensional spatial mapping in deeply annotated patient-derived primary and metastatic clinical samples. Towards this goal, our first aim will be to develop experimental design methods to select the patients, samples, and experimental parameters for creation of these three atlases. This aim is based on the rationale that atlas construction poses novel statistical challenges in experimental design that must be developed to maximally utilize resources towards atlas construction. In our second aim we will construct and implement the infrastructure of the data analysis unit at scale. The rationale for this aim is that quality control, and scalable quantification and annotation of the data at large scale will guide the use of the atlas for searching, comparing and interpreting samples. Finally, we will develop novel computational methods for data integration and interpretation towards a spatial-temporal atlas. In this instance we posit that data from different technologies and platforms often include redundant, biologically informative information that can be extracted for producing an annotated atlas. We will illustrate the impact of our atlas in the use case of predicting those early stage lung adenocarcinomas that are likely to metastasize to the brain. Ultimately the constructed atlases will provide insight on convergent events and bottlenecks in the metastatic transition, suggesting potential therapeutic targets and opportunities for intervention.