Stochastic Simulation Service: A Cloud Computing Framework for Modeling and Simul

描述（由申请人提供）：随机性在许多生物过程中起着重要作用。例子包括遗传学开关、噪声增强的振荡鲁棒性、波动增强的灵敏度或“随机聚焦”。许多细胞系统，包括发育，形态发生，极化和趋化性依赖于空间随机噪声的鲁棒性能。 同时，随机模拟是复杂的，消耗大量的计算机时间。它们可能要求研究人员能够熟练使用一个或多个复杂的软件包。学习使用现有的仿真工具并将其与其他软件集成需要相当长的时间。在许多情况下，这些工具并不存在，需要数学家和计算机科学家的专业知识来开发它们。通常，研究人员必须购买和维护计算机集群来执行大规模计算。所有这些都增加了研究过程的成本和延迟。 目前，还没有软件包可以让研究人员轻松构建生物系统的随机模型，并将其扩展到更高的细节和复杂程度。我们建议建立一个环境，建模者可以把他/她的注意力集中在生物学上;减轻软件安装和版本，数学算法，代码优化，计算机系统等的负担。这个环境将在笔记本电脑和计算机工作站上运行（对于小问题），按需扩展到高性能计算集群、网格和公共或私有云;从而为各种规模的模拟创造了一种经济高效的节能解决方案。我们将为这个环境配备最先进的软件，以解决关键类的问题，并使软件开发人员能够轻松集成新的和改进的算法，而无需开发自己的软件基础设施。我们将开发新的算法和软件，以解决以前无法实现的关键计算能力：（1）完全自适应的混合求解器STI（和nonSTI）良好混合系统（2）罕见事件概率的精确计算，以及（3）空间随机系统的模拟速度比以前的方法快几个数量级。这样一个社区资源的可用性将使生物学和算法开发的进展能够加速。

DESCRIPTION (provided by applicant): Stochasticity plays an important role in many biological processes. Examples include bistable genetic switches, noise enhanced robustness of oscillations, and uctuation enhanced sensitivity or "stochastic focusing". Numerous cellular systems, including development, morphogenesis, polarization and chemotaxis rely on spatial stochastic noise for robust performance. At the same time, stochastic simulations are complex and consume large amounts of computer time. They may require the researcher to be procient in the use of one or more complex software packages. Learning to use existing simulation tools and to integrate them with other software takes considerable time. In many cases, the tools do not exist and require the expertise of mathematicians and computer scientists to develop them. Often, researchers must purchase and maintain clusters of computers to perform the large-scale computations. All of this adds costs and delays to the research process. Currently, there exists no software package that allows researchers to easily build a stochastic model of a biological system, and scale it up to increasing levels of detail and complexity. We propose to build an environment where the modeler can focus his/her attention on the biology; alleviating the burden of software installation and versions, mathematical algorithms, code optimizations, computer systems, etc. This environment will run on laptops and computer workstations (for small problems), extending on demand to high-performance compute clusters, grids, and public or private clouds; thus creating a cost-eective and energy-ecient solution for simulations of all sizes. We will equip this environment with state of the art software for key classes of problems, and make it easy for software developers to integrate new and improved algorithms without the need to develop their own software infrastructure. We will develop new algorithms and software to address key computational capabilities that have not previously been attainable: (1) fully- adaptive, hybrid solvers for sti (and nonsti) well-mixed systems (2) ecient computation of probabilities of rare events, and (3) simulation of spatial stochastic systems at speeds that are several orders of magnitude faster than previous methods. The availability of such a community resource will enable and accelerate progress in both biology and algorithm development.