Computational Modelling of Neural Network Growth and Dynamics

许多脑部疾病是由神经发育的改变引起的；例如精神分裂症、自闭症和某些类型的癫痫。这些发育变化导致神经网络拓扑结构与健康受试者不同。例如，最近对脑电图（EEG）同步网络的研究发现，阿尔茨海默病、精神分裂症和癫痫患者的网络组织发生了特征性变化。为了了解这些疾病，有必要找出哪些发育因素导致网络拓扑结构改变并导致功能改变，如波或大规模激活（如癫痫发作期间）。由于体外实验通常是有限的，条件很难控制，我建议开发第一个神经发育的计算机模型，以测试假设和为未来的实验提供信息。通过高性能计算机模拟和网络科学和图论方法确定发育因素的作用，为计算神经解剖学的研究引入了一个新的方向。这些目标将通过以下两个目标来实现：(1)连接发育因子和网络拓扑：研究时间、空间布局和活动对生成的神经拓扑的作用。虽然网络生成的方法已经在网络科学领域进行了研究，但很少考虑到空间组织和时间，因此大多数模型对于生物系统来说是不现实的。(2)连接网络拓扑和动力学：将发育模拟产生的拓扑与实验可观察到的特征（波、延迟、振荡）联系起来。由于不是所有的拓扑结构都可以测试，我们将重点关注那些具有与视网膜相似特征的拓扑结构来观察波的传播，以及那些具有与皮层纤维束网络相似特性的拓扑结构来观察延迟和振荡。这些问题也是两大挑战的核心：理解大脑和思维的架构（GC5, UK Comp. Res. Comm.）和构建大脑（GC4， EPSRC网络“为英国微电子设计研究制定共同愿景”）。简单地观察神经组织可能不足以理解如何将这种组织转化为具有不同约束的技术系统。拟议的项目可以测试哪些开发配置和由此产生的网络拓扑结构导致等效动力学，从而可以确定导致可扩展和可行的技术设计的配置。对发育因素和网络行为之间关系的深入了解也将有助于解释发育性疾病，可能导致新的治疗方法。我们将与工程、发育生物学和制药研究方面的合作者讨论这些应用。

Many brain diseases are caused by changes of neural development; e.g. schizophrenia, autism, and certain kinds of epilepsy. These changes of development result in neural network topologies that differ from those of healthy subjects. Recent studies of EEG (electroencephalography) synchronization networks, for example, found characteristic changes in network organisation for Alzheimer, schizophrenia, and epilepsy patients. To understand these diseases, it is essential to find out which developmental factors lead to altered network topologies and resulting functional changes such as waves or large-scale activations (as during epileptic seizures). As in vitro experiments are often limited and conditions are hard to control, I propose to develop the first in silico model of neural development in order to test hypotheses and inform future experiments.This proposal introduces a new direction of research in computational Neuroanatomy by determining the role of developmental factors through high-performance computer simulations and methods from network science and graph theory. These aims will be reached through the following two objectives:(1) Linking developmental factor and network topology: Studying the role of timing, spatial layout, and activity on the generated neural topology. Whereas methods for network generation have been investigated in the field of network science, very few take into account the spatial organisation and timing and therefore most models are not realistic for biological systems.(2) Linking network topology and dynamics: Linking the topology yielded by developmental simulations to experimentally observable features (waves, latencies, oscillations). As not all topologies can be tested, we will focus those with similar characteristics as the retina observing wave propagation and those with similar properties to cortical fibre tract networks for observing latencies and oscillations. These questions are also at the core of two grand challenges: understanding the architecture of brain and mind (GC5, UK Comp. Res. Comm.) and Building Brains (GC4, EPSRC network 'Developing a Common Vision for UK research in Microelectronic Design'). Simply observing neural organisation might not be sufficient to understand how to translate this organisation to technical systems with different constraints. The proposed project can test which developmental configurations and resulting network topologies lead to equivalent dynamics and thus could identify configurations which lead to a scalable and feasible technical design. Insights into the relation between developmental factors and network behaviour will also help to explain developmental diseases potentially leading to new therapeutic treatments. We will discuss these applications with collaborators in engineering, developmental biology, and pharmaceutical research.