标题： 一种数据驱动的生物物理网络模型再现了秀丽线虫前运动神经动力学 显示英文标题

标题： A data-driven biophysical network model reproduces C. elegans premotor neural dynamics

摘要： 线虫的运动由向前和向后状态之间的转换组成，这些转换被转弯所打断。 这种运动能力对于线虫朝向吸引性刺激移动、逃避有害化学物质和探索其环境是必要的。 尽管实验人员已经确定了一些前运动神经元作为向前和向后运动的驱动因素，但这些神经元如何协同工作以产生观察到的行为仍需理解。 为了更好地理解线虫的神经动力学，我们在本文中提出了一种参数最少的生物物理动力系统模型，用于前运动网络。 我们的模型由一个由前运动神经元组成的循环连接集合（核心组）组成，该集合由超过一百个感觉和中间神经元驱动，这些神经元为核心组提供多样化的前馈输入。 该模型是数据驱动的，因为核心组中神经元的选择遵循实验指导，解剖结构由连接组决定，生理参数则从全脑成像和电压钳数据中推导出来。 当用现实的输入信号进行模拟时，我们的模型产生的前运动活动与实验数据非常相似：从看似随机的向前和向后行为之间的切换，到子网络与各种高阶统计量的同步。 我们认为间隙连接和突触连接在切换动力学中扮演着不同的角色。 由于接收到的突触输入，该模型正确预测了诸如停留与游荡等行为，并且我们证明它可以用来研究某些个体神经元的活动水平如何影响行为。 显示英文摘要

摘要： C. elegans locomotion is composed of switches between forward and reversal states punctuated by turns. This locomotory capability is necessary for the nematode to move towards attractive stimuli, escape noxious chemicals, and explore its environment. Although experimentalists have identified a number of premotor neurons as drivers of forward and reverse motion, how these neurons work together to produce the behaviors observed remains to be understood. Towards a better understanding of C. elegans neurodynamics, we present in this paper a minimally parameterized, biophysical dynamical systems model of the premotor network. Our model consists of a recurrently connected collection of premotor neurons (the core group) driven by over a hundred sensory and interneurons that provide diverse feedforward inputs to the core group. It is data-driven in the sense that the choice of neurons in the core group follows experimental guidance, anatomical structures are dictated by the connectome, and physiological parameters are deduced from whole-brain imaging and voltage clamps data. When simulated with realistic input signals, our model produces premotor activity that closely resembles experimental data: from the seemingly random switching between forward and reversal behaviors to the synchronization of subnetworks to various higher-order statistics. We posit that different roles are played by gap junctions and synaptic connections in switching dynamics. The model correctly predicts behavior such as dwelling versus roaming as a result of the synaptic inputs received, and we demonstrate that it can be used to study how the activity level of certain individual neurons impacts behavior.