UCSC-Buck Genome Data Analysis Center for the Genomic Data Analysis Network v2.0

摘要 肿瘤异质性--肿瘤亚克隆的复杂混合，最初转化的起源细胞， 肿瘤亚克隆在体内选择性压力和治疗过程中的演变及其相互作用 这些具有肿瘤微环境(TME)的细胞--促成了 肿瘤，是癌症进展和患者对治疗反应的关键决定因素。大规模 基因组学项目，如癌症基因组图谱(TCGA)和基因组数据分析网络(GDAN) 从不同肿瘤类型的多组学概述中揭示了重要的特征和模式。 然而，如何最大限度地利用这些数据来选择最佳路线仍然是一个谜 针对个别病人的治疗。拟议的GDAN将通过收集临床数据来弥合这一知识差距 与多个组学平台一起提供关键联系的信息和成果终端 用来训练有监督的计算方法。我们建议将我们的关键能力贡献给 途径分析，综合机器学习，mRNA-SEQ分析，驱动体细胞突变的评估， 以及高通量数据集的可视化，以服务于未来的GDAN分析工作组(AWG) 实现这些目标。我们将广泛收集和共享一个基因表达签名数据库，该数据库捕获 从大量的单细胞mRNA测序数据收集的细胞状态信息，例如从 人体细胞图谱(目标1)。此外，我们将贡献我们现有的和新的扩展， 机器学习方法，如AKIMATE，以最大限度地结合使用这些签名和其他签名 AWG批准的组学数据集作为功能，为GDAN的研究训练准确的反应预测因子，如 炼金术士(目标2)。我们的提案将调整TumorMap以使每周分析受益，并支持 探索和发布成果。具体地说，我们将与该小组合作创建新地图，以显示 分别对患者样本进行TME和TIC比较以帮助阐明新的重要亚型 由收集的数据所暗示的(目标3)。正如我们在过去12年为TCGA和GDAN所做的那样，我们 建议在这些努力中继续与财团密切合作，以显著丰富我们的 了解肿瘤异质性的分子和细胞基础及其对癌症的影响 进展和治疗反应。

ABSTRACT Tumor heterogeneity -- the complex mix of tumor subclones, the cell-of-origin that first became transformed, the evolution of tumor subclones under selective pressures of the body and due to treatment, and the interplay of these cells with the tumor microenvironment (TME) -- contributes to the character, behavior, and mystery of tumors and is a key determinant of cancer progression and a patient’s response to therapy. Large-scale genomics projects like the Cancer Genome Atlas (TCGA) and the Genome Data Analysis Network (GDAN) have revealed important characteristics and patterns from a multi-omics overview of various tumor types. However, it remains a mystery on how to maximize the use of these data to choose the best course of treatment for an individual patient. The proposed GDAN will close this gap in knowledge by collecting clinical information and outcomes endpoints alongside the multiple omics platforms that will provide key linkages upon which to train supervised computational approaches. We propose to contribute our key competencies of pathway analysis, integrative machine-learning, mRNA-seq analysis, assessment of driving somatic mutations, and visualization of high-throughput datasets to serve the future GDAN analysis working groups (AWGs) to achieve these goals. We will collect and share widely a database of gene expression signatures that capture cell state information gleaned from the large collection of single-cell mRNA sequencing data such as from the Human Cell Atlas (Aim 1). In addition, we will contribute our existing, and novel extensions to, machine-learning approaches like AKIMATE to maximally use these signatures and others in combination with AWG-approved omics datasets as features to train accurate predictors of response for the GDAN’s studies like ALCHEMIST (Aim 2). Our proposal will adapt the TumorMap to benefit weekly analysis and bolster the exploration and publication of results. Specifically, we will work with the group to create new maps that show the TME and TIC comparisons of the patient samples separately to help elucidate new important subtypes implied by the collected data (Aim 3). As we have done for the past twelve years for TCGA and the GDAN, we propose to continue working closely with the consortium in these endeavors to significantly enrich our understanding of the molecular and cellular basis of tumor heterogeneity and its influence on cancer progression and treatment response.