Theory of biochemical reaction networks in cells: understanding and exploiting stochastic fluctuations

通过分析随机波动来推断元件之间潜在的相互作用在物理学上有很长的历史。在生命科学中，最近的许多实验工作都集中在测量单细胞中的非遗传可变性，其中随机效应显着影响生化过程。因此，对分析和解释复杂生物系统中的随机波动的理论方法的需求是巨大的。可靠地将观察到的细胞间的可变性与潜在的分子相互作用联系起来，对于理解许多由随机效应形成的关键细胞过程是必要的，例如干细胞的分化、癌细胞对药物治疗的反应以及抗生素耐药性在细菌群体中的传播。 然而，由于生命系统不在热力学平衡下运行，我们不能使用像涨落-耗散-定理这样的一般关系，我们被迫单独研究每个生物过程。这意味着我们必须对所有组件之间的所有交互进行建模，要么猜测许多未知的细节，要么进行大致的近似。这种方法是不可靠的，即使在理论论文分析相同的实验数据时，也会导致相互矛盾的答案。 我们研究的目标是通过建立适用于整个系统类的通用属性来解决这一根本挑战，而不做大量显式或隐式的假设。我们的第一个主题重点是了解在复杂的生化反应网络中随机波动是如何产生、传播和消除的原理。我们将通过分析不同的“网络基序”如何塑造嵌入未指定反应网络中的单个组件的随机动力学来做到这一点。例如，我们将得出前馈循环、复杂形成和随机酶-底物相互作用的动力学所固有的基本限制和权衡，所有这些都嵌入到任意复杂的调控网络中。 在我们的第二个焦点中，我们认为随机波动不是理解生物过程的复杂因素，而是提取额外信息的机会。为了实现这一点，我们将开发一个理论框架，利用自然发生的随机波动作为非扰动工具来探索大型网络中的局部相互作用。这个框架的最终成功将是一种算法，它只根据观察到的成分的联合概率分布来生成建议的机制和分子反应的列表。除了开发新的理论工具外，我们还将进行实验合作，并应用我们的新方法来分析来自单细胞实验的高通量显微镜数据。

Analyzing stochastic fluctuations to infer underlying interactions of components has a long history in physics. In the life sciences much recent experimental work has focused on measuring non-genetic variability in single cells where stochastic effects significantly affect biochemical processes. As a result, there is an enormous demand for theoretical approaches to analyze and interpret stochastic fluctuations in complex biological systems. Reliably relating the observed cell-to-cell variability to underlying molecular interactions is necessary to understand many key cellular processes shaped by stochastic effects, such as the differentiation of stem cells, the response of cancer cells to drug treatment, and the spread of antibiotic resistance in bacterial populations. However, because living systems do not operate at thermodynamic equilibrium we cannot use general relations like the Fluctuation-Dissipation-Theorem and we are forced to study each biological process individually. That means we have to model all interactions between all the components either guessing many unknown details or making sweeping approximations. This approach is unreliable and has led to contradictory answers even when theoretical papers analyze the same experimental data. The goal of our research is to address this fundamental challenge by establishing universal properties that apply to entire classes of systems without making a large number of explicit or implicit assumptions. Our first thematic focus is to understand the principles of how stochastic fluctuations are generated, transmitted, and eliminated in complex biochemical reactions networks. We will do so by analyzing how different "network motifs" shape the stochastic dynamics of individual components embedded within unspecified reaction networks. For example, we will derive fundamental limits and trade-offs inherent to the dynamics of feed-forward loops, complex formation, and stochastic enzyme-substrate interactions, all embedded within arbitrarily complex regulatory networks. In our second focus we consider stochastic fluctuations not as a complication to understand biological processes but as an opportunity to extract additional information. To accomplish that we will develop a theoretical framework that exploits naturally occurring stochastic fluctuations as a non-perturbative tool to probe local interactions within large networks. The ultimate success of this framework will be an algorithm that produces a list of suggested mechanisms and molecular reactions based solely on the observed joint probability distributions of components. In addition to developing new theoretical tools we will pursue experimental collaborations and apply our new methods to analyze high-throughput microscopy data from single-cell experiments.