Data-driven agent-based modelling of Trypanosoma collective behaviour

单细胞寄生虫布鲁氏锥虫在采采蝇中肠呈线性排列。在琼脂糖凝胶上进行的体外社会运动测定中，虫形寄生虫的菌落显示出放射状突起。集体行为的定量特征是可用的，但潜在的机制以及与体内情况的联系需要进一步调查。我们通过数据驱动的基于主体的集体行为模型的开发来解决这些问题，该模型来源于Vicsek模型（Vicsek et al. 1995）。考虑到物理以及化学试剂-试剂和试剂-边界的相互作用，我们测试了我们的主要假设，即锥虫的集体运动可以通过负自趋化性和寄生虫在边界的对齐的组合来复制。该模型在编程语言Julia中的计算效率实现允许模拟与实验中寄生虫数量相当的代理数量。因此，将模拟结果与实验数据进行直接、定量的比较是可行的。我们考虑了文献中的数据以及优先计划中其他项目的数据。与实验项目的合作可以评估生物学相关性。通过与开发寄生虫的详细流体动力学模拟的项目相互作用来测试模型在物理原理近似方面的质量。在建立了准二维体外实验的适当计算表示之后，我们将代理转移到三个空间维度，以建立与体内情况的联系。特别地，我们考虑一个通道与流场以及收缩和障碍来代表蝇肠。在这种情况下分析我们的模型提供了对锥虫集体行为对环境几何形状变化的敏感性的见解。在基于智能体的模型的开发过程中，我们强调了通用性。为了形成更广泛采用我们的模型的基础，我们在优先计划中确定了三个拟议的项目，以调查大群体个体的运动模式和相互作用，即小肠中的多聚螺旋虫，小肠中的贾第鞭毛虫和液滴中的恶性疟原虫微管蛋白二聚体。在这些系统上测试我们的建模方法有助于我们进一步理解寄生虫学中的一般物理概念。我们的项目有助于更详细地了解锥虫的运动和微环境中与边界相互作用的物理学。此外，它还促进了其他寄生系统的类似研究。

The unicellular parasite Trypanosoma brucei exhibits swarming with linear alignment in the tsetse fly midgut. Colonies of the insect-form of the parasite show radial projections in in vitro social motility assays on agarose gel. Quantitative characterisations of the collective behaviour are available, but the underlying mechanisms as well as the link to the in vivo situation require further investigations. We address these questions by data-driven development of an agent-based model for collective behaviour derived from the Vicsek model (Vicsek et al. 1995). Considering physical as well as chemical agent-agent and agent-boundary interactions, we test our main hypothesis that collective motion of trypanosomes can be reproduced by a combination of negative auto-chemotaxis and parasite alignment at the boundary. A computationally efficient implementation of the model in the programming language Julia allows simulations for agent numbers comparable to the number of parasites in the experiments. Hence, a direct, quantitative comparison of the simulation results to experimental data is feasible. We consider both data from the literature as well as from other projects in the Priority Programme. Collaboration with experimental projects allows assessing the biological relevance. The model quality in terms of the approximation of physical pronciples is tested through interaction with projects that develop detailed hydrodynamic simulations of the parasite. Having established an appropriate computational representation of the quasi two-dimensional in vitro assay, we transfer our agents to three spatial dimensions to establish a link to the in vivo situation. In particular, we consider a channel with a flow field as well as constrictions and obstacles to represent the fly gut. Analysis of our model in this setting provides insight into the sensitivity of the collective behaviour of trypanosomes to changes of the environment geometry. During the development of the agent-based model, we place an emphasis on generality. To form the basis for a wider adoption of our model, we have identified three proposed projects in the Priority Programme that investigate the movement patterns and interactions of large groups of individuals, namely Heligmosomoides poligyrus in the small intestine, Giardia muris in the small intestine and Plasmodium falciparum tubulin dimers in liquid droplets. Testing our modelling approach on these systems helps to further our understanding of general physical concepts in parasitology. Our project contributes to a more detailed understanding of Trypanosoma locomotion and the physics of the interaction with boundaries in the microenvironment. In addition, it promotes similar studies in other parasitic systems.