Deep Curation via an Integrated Whole-Cell Computational Model

研究摘要/摘要 生物数据的产生正在迅速地为我们提供最苛刻的数据分析之一 世界面临的前所未有的挑战--不仅是在存储和可访问性方面，也许更关键的是 就其广泛的异质性和变异性而言。在这份提案中，我们提出了一种新的方法来解决这些问题 挑战，我们称之为“深冷”：一种大规模、集成的建模方法，可以同时 对数百万个不同类型的数据进行交叉评估。“深”一词反映了多重 我们执行的管理层，不仅包括数据的层，还包括从这些数据派生的参数的层 数据、数学方程、统一模型和仿真输出。因此，经过深度策划的 模型是自动处理、整理和分析数据的无价工具。我们计划在以下方面作出努力 深度冷冻是基于大肠杆菌的计算机模型，该模型解释了粗略的 40%的注释良好的基因，并基于一系列广泛的不同测量汇编而成 数以千计的报告(目前在《科学》杂志的第二轮审查中)。这项提议的目标是扩大这一点 支持深度修复与当前未合并的100个环境中的增长相关的数据的模型。我们 然后，可以作为统一的整体同时评估数据集的交叉一致性，确定关键的 数据集不是交叉一致的区域，因此需要进一步的实验研究。 这一提议的意义在于，深冷代表着我们在 能够利用大量异质、可变和复杂的生物数据集；它自动化和 加速了变革性的生物医学发现；我们将在EcoCyc、 关于任何生物体的最全面的数据库，以及现存最复杂的生物模型；以及 全细胞建模是一个快速增长的领域，随着它向更多 复杂的细胞和细胞群。这一提议的创新之处在于，深冷是一种 一种全新的、高度创新的方法，这是目前世界上任何其他实验室都无法获得的； 拟议的工作将产生一个前所未有的复杂性的显著扩展的全细胞模型； 作为新颖和高度创新的建模技术；我们包括关于以下方面的显式知识精选 除数据外的机制；以及EcoCyc数据库和 大肠杆菌模型将以协同的方式极大地扩大两者的容量、范围和可见性。我们的 具体目标是：目标1(持续)，建立与不同种类的大肠杆菌生长相关的数据和参数层 环境；目标2(建模)，实现方程、模型和模拟层；目标3(深冷)， 使用综合模型在全生物体范围内交叉评估统一数据集；以及目标4 (分发)，通过GitHub(软件工具)、EcoCyc(数据)将模型提供给更广泛的社区 和参数)和Google Cloud(模拟和交互可视化)。

Research Summary/Abstract The generation of biological data is rapidly presenting us with one of the most demanding data analysis challenges the world has ever faced - not only in terms of storage and accessibility, but perhaps more critically in terms of its extensive heterogeneity and variability. In this proposal, we present a new approach to these challenges, which we call “Deep Curation”: a large-scale, integrated modeling approach to simultaneously cross-evaluate millions of heterogeneous data against themselves. The word “deep” reflects the multiple layers of curation we perform, including layers not only for data, but also for parameters derived from these data, the mathematical equations, the unified model, and the simulation output. Thus, the deeply-curated model is an invaluable tool for processing, curating and analyzing data automatically. Our proposed efforts in Deep Curation are based on a computer model of Escherichia coli that accounts for the function of roughly 40% of the well-annotated genes, and is based on an extensive set of diverse measurements compiled from thousands of reports (currently in 2nd round of review at Science). The goal of this proposal is to expand this model to enable Deep Curation of data related to growth on >100 currently-unincorporated environments. We can then assess the cross-consistency of the data sets simultaneously, as a unified whole, identifying critical areas in which datasets are not cross-consistent and therefore further experimental investigation is needed. The Significance of this proposal is that Deep Curation represents a first-in-kind quantum leap forward in our ability to exploit massively heterogeneous, variable and complex biological datasets; that it automates and accelerates transformative biomedical discovery; that we will create a bi-directional pipeline between EcoCyc, the most comprehensive database on any organism, and the most complex biological model in existence; and that whole-cell modeling is a rapidly-growing field with transformative potential as it advances towards more complex cells and groups of cells. The Innovation associated with this proposal is that Deep Curation is a brand-new and highly innovative approach that is not currently available to any other lab in the world; that the proposed work will produce a dramatically expanded whole-cell model of previously-unseen complexity; as well as novel and highly innovative modeling technology; that we include explicit curation of knowledge regarding mechanism in addition to data; and that the automated communication between the EcoCyc database and the E. coli model will dramatically expand the capacity, scope and visibility of both in a synergistic way. Our Specific Aims are: Aim 1 (Curation), build the Data and Parameter layers related to E. coli growth on diverse environments; Aim 2 (Modeling), implement the Equation, Model and Simulation layers; Aim 3 (Deep Curation), use the integrated model to cross-evaluate the unified data set at the whole-organism scale; and Aim 4 (Distribution), make the model available to the broader community via GitHub (software tools), EcoCyc (data and parameters), and Google Cloud (simulations and interactive visualizations).