Integrating Multi-Scale Imaging, Reaction-Diffusion Simulation, and Markov Model Inference to Enhance Predictive Design and Interpretation of Single-Molecule Gene Regulation Experiments

项目摘要 单细胞成像可以量化基因调控的复杂时空动态， 重要的生物医学过程，从细菌感染到癌症。这种基因调控受到 生物过程的复杂性和随机性，其观察进一步受到测量 由于生物化学标记的低效和显微镜成像的失真而产生的伪影。然而，尽管这些 复杂性，初步工作表明，有可能整合数据和计算模型来预测 在无数的环境和遗传条件下的基因调控提供了：（1）模型必须受到限制 通过提供信息和可重复的数据，（2）模型必须严格验证，以说明生物和 技术上的变化，以及（3）必须系统地探讨模型，以量化不确定性。总体 该项目假设是，在亚细胞动力学中观察到的空间和时间波动包含 独特的信息，可以通过改进的计算方法和模型引导的实验来解锁。 为了验证这一假设，该项目将创建一个新的研究平台，称为单细胞图形 解释和设计实验的工具。scGUIDE将结合联合收割机实验分析（例如，图像处理 和单颗粒跟踪以从荧光显微镜实验中提取定量数据），空间 随机模拟（例如，反应-扩散模型以生成模拟细胞实验的真实视频）， 模型抽象和识别（例如，参数推断和不确定性量化 将定量观察转化为预测洞察力），以及实验设计（例如，统计方法来查明 哪些特定的实验条件最有可能揭示新的生物学见解）。 为了展示其广泛的功能，scGUIDE将用于分析和设计单细胞 四个不同的健康相关过程的实验。在酵母，该项目将审查协调 应激激活的MAPK动力学和Spt-Ada-Gcn 5乙酰转移酶（佐贺）亚基之间的关系， 染色质和RNA转录/转运动力学，并已牵连到癌，骨骼肌， 发育不良和视网膜变性。在人类细胞中，该项目将研究时空聚类， RNAP聚合酶II的磷酸化，因为它在CDK抑制剂癌症下参与单基因转录 治疗。在骨肉瘤细胞中，该项目将探索对局部tRNA资源的竞争如何影响 在不同的亚细胞区域和在人类和病毒环境中的单个mRNA分子的翻译。最后， 该项目将探索表观遗传记忆和分子竞争对多基因组的影响， pap操纵子的世代激活或抑制，使E.大肠杆菌建立尿路感染。 每个项目将建立机制和定量预测模型，以了解空间和时间的相互作用 转录或翻译因子、酶和复杂的分子机器联合收割机与环境 影响调节单基因，单mRNA，单细胞和群体水平的表达。

Project Summary Single-cell imaging can quantify intricate spatial and temporal dynamics of gene regulation that underly important biomedical process ranging from bacterial infections to cancer. This gene regulation is subject to complexities and randomness of biological processes, and its observation is further subject to measurement artifacts due to inefficiencies in biochemical labels and distortions in microscope imaging. Yet, despite these complications, preliminary work shows that it is possible to integrate data and computational models to predict gene regulation in myriad environmental and genetic conditions provided that: (1) models must be constrained by informative and reproducible data, (2) models must be rigorously verified to account for biological and technical variations, and (3) models must be systematically explored to quantify uncertainties. The overarching hypothesis of this project is that spatial and temporal fluctuations observed in subcellular dynamics contain unique information that can be unlocked with improved computational methods and model-guided experiments. To test this hypothesis, this project will create a new research platform to be known as the single-cell Graphical Utility to Interpret and Design Experiments. scGUIDE will combine experimental analysis (e.g., image processing and single-particle tracking to extract quantitative data from fluorescence microscopy experiments), spatial stochastic simulation (e.g., reaction-diffusion models to generate realistic videos to mimic cellular experiments), model abstraction and identification (e.g., parameter inference and uncertainty quantification to translate quantitative observations into predictive insight), and experiment design (e.g., statistical methods to pinpoint which specific experimental conditions are most likely to reveal new biological insight). To demonstrate its broad capabilities, scGUIDE will be used to analyze and design single-cell experiments for four different health-related processes. In yeast, the project will examine the coordination between stress-activated MAPK dynamics and Spt-Ada-Gcn5 Acetyltransferase (SAGA) subunits that control chromatin and RNA transcription/transport dynamics, and which have been implicated in carcinoma, skeletal dysplasia, and retinal degeneration. In human cells, the project will examine the spatiotemporal clustering and phosphorylation of RNAP Polymerase II as it engages in single-gene transcription under CDK-inhibitor cancer treatments. In osteosarcoma cells, the project will explore how competition for local tRNA resources affects translation of single-mRNA molecules in different sub-cellular regions and in human and viral contexts. Finally, the project will explore the effects that epigenetic memory and molecular competition have on the multi- generational activation or repression of the pap operon that allows E. coli to establish uropathogenic infections. Each project will build mechanistic and quantitatively predictive models for how spatial and temporal interactions of transcription or translation factors, enzymes, and complex molecular machines combine with environmental influences to regulate expression at the single-gene, single-mRNA, single-cell, and population levels.