DISTRIBUTED SIMULATION AND OPTIMIZATION OF GENE NETWORK MODELS

基因网络模型的分布式仿真与优化

• 批准号: 7956340
• 负责人: Jean M Peccoud
• 金额: $ 0.08万
• 依托单位: CARNEGIE-MELLON UNIVERSITY
• 依托单位国家: 美国
• 财政年份: 2009
• 资助国家: 美国
• 起止时间: 2009-08-01 至 2010-07-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/7956340
• 关键词: Address; Algorithms; Biomedical Research; Client; Communities; Computer Retrieval of Information on Scientific Projects Database; Data; Development; Environment; Equation; Funding; Genes; Grant; High Performance Computing; Hour; Institution; Manuscripts; Methods; Modeling; Research; Research Personnel; Resources; Running; Simulate; Solutions; Source; Stochastic Processes; United States National Institutes of Health; Work; biological systems; chemical kinetics; cost; network models; simulation

This subproject is one of many research subprojects utilizing the resources provided by a Center grant funded by NIH/NCRR. The subproject and investigator (PI) may have received primary funding from another NIH source, and thus could be represented in other CRISP entries. The institution listed is for the Center, which is not necessarily the institution for the investigator. Our group develops computational and experimental methods to estimate the parameters of gene network models. These models are generalization of chemical kinetics models. They can either represented by differential equations or by stochastic processes. We use optimization algorithms to fit gene network models to experimental data. The size of the parameter space of our models along with their nonlinearity can make them difficult to optimize. In situation where we have to use stochastic models to simulate the biological system that we have characterized experimentally, the computational cost of optimizing the model is even greater. Depending on the models and parameters used, a typical optimization problem can require 100s of CPU hours. In addition to getting us access to a HPC environment for our own use, I want to better understand the resources available through the TeraGrid that could be used to help bring the benefit of HPC to my scientific community. This is especially important because very few people in my field have the expertise to work in a HPC environment even though many people could benefit from it. The attached manuscript describes an application that we are about to release in which we have addressed this need by providing users with a build-in access to a cluster. Users run the application on their desktop and can elect to execute simulations on a remote machine. We have implemented this mechanism on a 8 node dedicated cluster as proof of concept but this solution is not scalable. We would like to see to what extent we could provide people in our community easy access to the TeraGrid through the development of specific client-server applications to analyze gene networks. Once we have a better understanding of the TeraGrid environment, I will submit a separate request for advanced support to analyze with the TeraGrid staff the possibility to develop a specific gateway. I dont have a specific grant associated with that request at this point. However, I need to be able to articulate how the TeraGrid fits in the research plans of many of several grants that I am currently preparing for the NIH and NSF.

该副本是利用众多研究子项目之一 由NIH/NCRR资助的中心赠款提供的资源。子弹和 调查员（PI）可能已经从其他NIH来源获得了主要资金， 因此可以在其他清晰的条目中代表。列出的机构是 对于中心，这不一定是调查员的机构。 我们的小组开发了计算和实验方法来估计基因网络模型的参数。这些模型是化学动力学模型的概括。它们可以由微分方程代表，也可以由随机过程表示。我们使用优化算法将基因网络模型拟合到实验数据。我们模型的参数空间的大小以及它们的非线性可能会使它们难以优化。在我们必须使用随机模型来模拟我们实验表征的生物系统的情况下，优化模型的计算成本更大。根据所使用的模型和参数，典型的优化问题可能需要100个CPU小时。除了让我们访问HPC环境以供自己使用，我还想更好地了解通过Teragrid可用的资源，这些资源可用于帮助将HPC的好处带给我的科学界。这一点尤其重要，因为我领域中很少有人有专业知识可以在HPC环境中工作，即使许多人可以从中受益。附件的手稿描述了我们将要发布的应用程序，通过为用户提供对群集的构建访问，以解决这种需求。用户在其桌面上运行应用程序，并可以选择在远程计算机上执行模拟。我们已经在8个节点专用群集作为概念证明上实现了这种机制，但该解决方案不可伸缩。我们希望通过开发特定的客户服务器应用程序来分析基因网络，可以在多大程度上让社区中的人们轻松访问Teragrid。一旦我们对Teragrid环境有了更好的了解，我将提交单独的高级支持请求，以与Teragrid员工分析开发特定网关的可能性。目前，我没有与该请求相关的特定赠款。但是，我需要能够阐明我目前正在为NIH和NSF做准备的几项赠款的研究计划中如何适应。