Tools for Large-Scale Analysis of Driver Pathways

描述(由申请人提供)： 癌症基因组图谱计划(TCGA)正在产生大量的遗传和表观遗传信息，这些信息有望阐明癌症的生物学机制，并为更有效的治疗指明方向。最新证据表明，在一个典型的肿瘤中，数百个基因的功能会因表达不足和过度表达、点突变、易位、异常拷贝数和甲基化模式而改变。与此同时，越来越清楚的是，这种复杂性可以通过按通路--特别是信号通路--组织与癌症相关的改变的数据来降低。以通路为中心的分析已经在差异基因表达、体细胞突变和拷贝数的研究中发挥了重要作用。在我们看来，通路方法代表着最有希望的方法，既可以理解癌症相关变化的复杂性，也可以快速将多平台的癌症基因组研究转化为预防和治疗方面的切实进展。 这个研究团队在癌症基因组学和特定的基因集分析方面拥有丰富的经验。我们最近开发了比较浓缩分析工具--调查几种相关的浓缩分析是否识别共同或不同的一组基因；比较网络分析--调查癌症和正常组织中途径成员之间的相互联系是否不同；我们积极参与与癌症基因组测序项目相关的以途径为中心的分析工具的开发。我们在这里建议通过开发多维工具来扩展这些努力，这些工具可以利用现有的路径信息来集成不同的数据类型。我们的工作将产生一个强大的分析工具，用于从TCGA试点项目即将提供的多维、高维数据集中提取生物相关性。 这项为期两年的提案的具体目标如下：目标1.算法。我们将在TCGA考虑的所有平台上开发基于途径的癌症相关改变分析的方法和算法。目标2.验证。我们将使用现有的TCGA数据来验证我们的算法，既可以执行分析，也可以生成真实的合成数据集，在这些数据集中，已知在癌症中起因果作用的途径和基因。我们将使用这些来系统地探索在受控环境中考虑的替代算法的性能。AIM 3.符合caBIG的实施。我们将开发适合在超级计算机集群上令人尴尬地并行执行的实现。这种方法将确保算法可扩展到涉及数百个基因组的数据集。我们将把我们的算法编码到caBIG银级兼容工具中，这些工具利用caBIG基础设施，包括通用数据元素。 考虑到直接“垂直”整合不同基因组测量产生的信息的巨大挑战，以路径为中心的分析方法是快速解释和临床翻译TCGA数据的最有前途的场所。如果成功，该项目将通过提供工具来确定哪些途径对个别患者的进展和预后重要，从而有助于加快个性化癌症治疗的发展。

DESCRIPTION (provided by applicant): The Cancer Genome Atlas Project (TCGA) is generating large amounts of genetic and epigenetic information that promise to illuminate the biological mechanisms underlying cancer and to point the way towards more effective treatments. The latest evidence indicates that that the functions of hundreds of genes are altered in a typical tumor by under- and over-expression, point mutation, translocation, aberrant copy number and methylation patterns. At the same time, it is becoming increasingly clear that this complexity can be reduced through organizing data about cancer-related alterations by pathways - particularly signaling pathways. Pathway-centric analysis already plays an important role in studies of differential gene expression, somatic mutations, and copy number. In our view, the pathway approach represents the most promising methodology for both understanding the complexity of cancer-related alterations, and for rapidly translating multi-platform genomic investigations of cancer into tangible progress in prevention and treatment. This investigative team has extensive experience in cancer genomics in general and gene set analyses specifically. We recently developed tools for comparative enrichment analysis - the investigation of whether several related enrichment analyses identify common or different sets of genes; comparative network analysis - the investigation of whether the interconnections between members of a pathway are different in cancer and normal tissues; and we contributed actively to tool development for pathway-centric analyses associated with cancer genome sequencing projects. We propose here to extend these efforts by developing multidimensional tools that can leverage existing pathway information to integrate across different data types. Our work will produce a powerful analytic tool for extracting biological relevance from the multiple, high-dimensional datasets that will soon be available from the TCGA Pilot Project. The specific aims of this two year proposal are the following: AIM 1. ALGORITHMS. We will develop approaches and algorithms for pathway-based analysis of cancer- related alterations across all the platforms considered in the TCGA. AIM 2. VALIDATION. We will validate our algorithms by using existing TCGA data, both to perform analyses and to generate realistic synthetic datasets in which the pathways and genes that play a causal role in cancer are known. We will use these to systematically explore the performance of the alternative algorithms considered in a controlled environment. AIM 3. caBIG COMPLIANT IMPLEMENTATION. We will develop implementations suitable for embarrassingly parallel execution on supercomputer clusters. This approach will ensure that the algorithms are scalable to datasets involving hundreds of genomes. We will code our algorithms into caBIG silver-level compatible tools that leverage caBIG infrastructure, including the common data elements. Given the significant challenges of direct "vertical" integration of information generated by different genome-wide measurements, pathway-centric analytic methods represent the most promising venue for fast interpretation and clinical translation of TCGA data. If successful, the project will help accelerate development of personalized cancer treatment, by providing tools to identify which pathways are important for progression and prognosis in individual patients.