CG-GRID: Computational Genetics Grid Resource for Interaction Discovery

描述（由申请人提供）：对癌症和心血管疾病等常见人类疾病的易感性是由许多遗传因素决定的，这些遗传因素在个人年龄和环境暴露的背景下以非线性方式相互作用。这种复杂的遗传结构对利用全基因组关联研究（GWAS）鉴定易感基因具有重要意义。虽然新的方法可用于模拟基因-基因和基因-环境相互作用，但在全基因组范围内对所有snp组合进行详尽的测试是不可实现的，因为比较的数量实际上是无限的。因此，我们开发识别和建模snp组合的智能策略是至关重要的。网格计算系统和随机搜索算法的最新进展能够适应大型组合问题，现在为执行GWAS交互分析提供了基础，这些分析以前被认为是难以处理的。该研究计划的目标是将多因素降维（MDR）方法与Parabon(R) Crush随机搜索程序（一种为大规模并行网格设计的机会进化搜索算法）的功能相结合，以模拟非线性基因-基因相互作用（Aim 1）。这两种算法的结合将使在全基因组范围内识别非线性基因-基因相互作用成为可能。通过利用Parabon计算网格（一种由互联网上数千台计算机提供支持的代理计算服务）的力量，这将使计算变得可行。我们将使用模拟全基因组关联数据（目标2）评估这些新算法，然后开发一个web界面和相应的服务，用于从浏览器启动Crush-MDR分析（目标3）。最后，我们将使用公开可用的全基因组关联数据演示Crush-MDR分析服务（目标4）。这些具体目标的实现将使Parabon进入第二阶段的STTR应用程序，通过修改Crush-MDR算法来扩展这项服务，使用统计和生物专家知识来源来增强对基因-基因相互作用复杂模式的搜索。其业务目标是向市场提供三种产品：基于Parabon计算网格的Crush-MDR，作为全基因组关联数据的基因-基因相互作用分析的按次付费服务；一个由Parabon发现的基因-基因相互作用的知识库，将授权给制药行业；以及基于使用Crush-MDR算法发现的知识，用于预测人类疾病终点的专有基因分型分析。这里提出的具体目标将通过向研究界提供基于网络的“软件即服务”（SaaS）交付模型中的强大分析算法和高性能计算基础设施来改变人类疾病的遗传分析，这种模式易于访问、负担得起且易于使用。更重要的是，最终产生的遗传知识和分析将有助于彻底改变基因组信息的使用，以改善人类的医疗状况。

DESCRIPTION (provided by applicant): Susceptibility to common human diseases such as cancer and cardiovascular disease is determined by numerous genetic factors that interact in a nonlinear manner in the context of an individual's age and environmental exposure. This complex genetic architecture has important implications for the use of genome- wide association studies (GWAS) for identifying susceptibility genes. While novel methods are available for modeling gene-gene and gene-environment interactions, exhaustive testing of all combinations of SNPs is not feasible on a genome-wide scale because the number of comparisons is effectively infinite. Thus, it is critical that we develop intelligent strategies for identifying and modeling combinations of SNPs. Recent advances in grid computing systems and stochastic search algorithms able to accommodate large combinatorial problems now provide a basis for performing GWAS interaction analyses that were previously thought to be intractable. The objective of this research program is to combine the multifactor dimensionality reduction (MDR) approach to modeling nonlinear gene-gene interactions with the power of the Parabon(R) Crush" stochastic search procedure - an opportunistic evolutionary search algorithm designed for massively parallel grids (Aim 1). The combination of these two algorithms will make it feasible to identify nonlinear gene-gene interactions on a genome-wide scale. This will be made computationally feasible by harnessing the power of the Parabon Computation Grid - a brokered computation service powered by thousands of computers across the Internet. We will evaluate these new algorithms using simulated genome-wide association data (Aim 2) and then develop a web interface and corresponding service for launching Crush-MDR analyses from a browser (Aim 3). Finally, we will demonstrate the Crush-MDR analysis service using publicly available genome-wide association data (Aim 4). Accomplishment of these specific aims will position Parabon for a Phase II STTR application that will expand this service by modifying the Crush-MDR algorithm to use both statistical and biological sources of expert knowledge to enhance the search for complex patterns of gene-gene interactions. The business objectives are to bring to market three offerings: Crush-MDR atop the Parabon Computation Grid as a pay-per-use service for gene-gene interaction analysis of genome-wide association data; A knowledge base of gene-gene interactions discovered by Parabon that will be licensed to the Pharmaceutical industry; and Proprietary genotyping assays for predicting human disease endpoints that are based on knowledge discovered using the Crush-MDR algorithm. The specific aims proposed here will transform the genetic analysis of human diseases by making available to the research community powerful analysis algorithms and high-performance computing infrastructure in a web-based "Software as a Service" (SaaS) delivery model that is accessible, affordable and easy to use. More importantly, the genetic knowledge and assays that ultimately result will help revolutionize the use of genomic information to improve the human medical condition. PUBLIC HEALTH RELEVANCE: Susceptibility to common human diseases such as cancer and cardiovascular disease is determined by numerous genetic factors that interact in a nonlinear manner in the context of an individual's age and environmental exposure. The goal of this research program is to develop the computer software and computing technology to enable to researchers to identify combinations of genetic and environmental factors that are associated with human health and disease.