StochSS: A Next-Generation Toolkit for Simulation-Driven Biological Discovery

项目摘要 数学模型的开发对于理解复杂的生物过程至关重要，因为它将当前的理解编成法典，以便可以根据现有的数据进行测试。一个好榜样 有足够的细节可以用来确定潜在的干预点(例如，药物靶点) 这一过程的不受欢迎的结果(例如，疾病的影响)可能被改变。模型开发通过模型构建、模型在多种条件下的模拟以及 与实验数据进行了比较。这个循环是重复的，并经常通过新的实验来扩大 更多的数据，直到得到的模型可以可信地解释数据。大量的时间和时间 必须致力于寻找和/或开发工具来分析模型并将其与数据进行比较， 拟合参数并评估数据和数据中典型的大量不确定性的影响 参数，模拟模型并分析模拟数据，改进模型以更好地捕捉我们的 在过程的每个阶段增加了解，确定哪些额外的实验将增加最多 根据我们的理解，我们在拟议工作中的目标是促进和加速建模 通过提供最先进的、集成良好的工具来报告完整和信息丰富的每个结果的流程 阶段，使建模者和实验者能够专注于他们最擅长的事情：科学发现。 这是一项更新建议，它建立在当前 项目。在这项工作中，我们开发了StochSS，这是一种用于定量建模的新型软件即服务产品 能够在公共云环境中无缝部署的生化网络。StochSS做了一个 出色地支持建模过程的两个主要步骤：模型构建-采取您的 模型描述，并将其转换为StochSS模拟引擎可以使用的形式，以及 模拟-执行模拟以产生结果。 拟议的项目有三个相辅相成的目标。首先是进一步开发StochSS的核心 能力，并采取步骤确保其长期可持续性；第二是制定一项 一是开发模型开发工具包，三是开发模型探索工具包。这两个工具包 将集成到我们现有的StochSS模型构建和仿真环境中，并将利用 我们现有的云计算软件基础设施。 目标1.核心能力和长期可持续性这一目标有三个分目标：(1)建立 有助于确保社区参与和更好的StochSS长期可持续性的实践 NIH资金，(2)扩展核心StochSS功能能力，以及(3)改善与其他 软件通过对标准数据格式的支持。 目标2.模型开发工具包开发和集成工具，以促进和加速这一进程 模型开发：通常需要的(建模、模拟、实验)的迭代 采用最合理的模型来解释数据。模型开发工具包将 解决不确定度的参数估计和量化、生成和评估 合理的模型，以及实验的最优设计。 目标3.模型探索工具包为模型探索开发和集成工具：过程 探索参数空间，以确保模型对不确定和/或 未确定的参数，以找到参数空间的区域，在该区域中模型能够产生 给定的行为，并发现该模型能够在 不确定和/或不确定参数的空间。

Project Summary The development of a mathematical model is critical to the understanding of complex biological processes because it codifies current understanding so that it can be tested against existing data. A good model with sufficient detail can be used to identify potential points of intervention (for example, drug targets) at which an undesired outcome (for example, effects of disease) of the process might be altered. Model development proceeds through a cycle of model building, simulation of the model under numerous conditions, and comparison to experimental data. The cycle is repeated and often augmented by new experiments to capture additional data, until the resulting model can plausibly explain the data. Tremendous amounts of time and effort must be devoted to finding and/or developing tools to analyze the model and compare it to the data, fit the parameters and assess the effects of typically large amounts of uncertainty in both the data and the parameters, simulate the model and analyze the simulation data, refine the model to better capture our increased understanding at each stage of the process, decide which additional experiments would add most to our understanding, etc. Our objective in the proposed work is to facilitate and accelerate the modeling process by providing state of the art, well-integrated tools to report complete and informative results at each stage, enabling the modeler and the experimentalist to focus on what they do best: scientific discovery. This is a renewal proposal that builds on the capabilities and infrastructure developed in the current project. In that work we developed StochSS, a novel Software-as-a-Service offering for quantitative modeling of biochemical networks capable of seamless deployment in public cloud environments. StochSS does an excellent job of supporting two of the major steps of the modeling process: Model Building - taking your model description and putting it into a form that the StochSS simulation engines can work with, and Simulation - performing the simulations to produce the results. The proposed project has three complementary Aims. The first is to further develop StochSS's core capabilities and to take the steps that will ensure its long-term sustainability; the second is to develop a Model Development Toolkit, and the third is to develop a Model Exploration Toolkit. Both of these toolkits will be integrated into our existing StochSS Model Building and Simulation environment and will leverage our existing software infrastructure for cloud computing. Aim 1. Core Capabilities and Long-Term Sustainability This aim has three sub-aims: (1) instituting practices that will help ensure community involvement and better long-term sustainability of StochSS beyond NIH funding, (2) extending core StochSS functional capabilities, and (3) improving compatibility with other software via support for standard data formats. Aim 2. Model Development Toolkit Develop and integrate tools to facilitate and accelerate the process of Model Development: the iterations of (modeling, simulation, experiment) that are typically required to converge on the most plausible model that can explain the data. The Model Development Toolkit will address parameter estimation and quantification of uncertainty, generation and evaluation of the set of plausible models, and optimal design of experiments. Aim 3. Model Exploration Toolkit Develop and integrate tools for Model Exploration: the process of exploring the parameter space to ensure that the model is robust to variations in uncertain and/or undetermined parameters, to find the regions of parameter space in which the model is capable of yielding a given behavior, and to discover all of the qualitatively distinct behaviors that the model is capable of within the space of uncertain and/or undetermined parameters.