Modelling and Simulation of Granuloma Formation in Visceral Leishmaniasis Anton Jakob Fl¨ ugge * , Jon Timmis *† , Paul Andrews * , John Moore ‡ and Paul Kaye ‡ * Department of Computer Science, University of York, Heslington, York, UK Email: jtimmis@cs.york.ac.uk † Department of Electronics, University of York, Heslington, York, UK ‡ Center for Infection and Immunology, University of York, Heslington, York, UK Abstract— Visceral leishmaniasis is a parasitic disease that can cause death. It is characterized by the formation of granuloma structures in the liver that can form at different time points after the infection. To date, the possible processes underlying granuloma formation are not fully understood. The importance of modelling in immunology is increasing, as many immunologucal phenonema are hard to study in vivo over periods of time, and hence modelling can provide some insight that might help deepen the understanding of the phenonema or help guide experimental work. This paper discusses initial studies into the formation of granuloma using a combination of UML like modelling and agent based simulation. I. I NTRODUCTION In a recent paper it has been argued that the area of Artificial Immune Systems (AIS) is widening to encompass the area of immunological modelling [1]. Along with [2] and [3] they argue for a greater interaction between computer science and the immunological community, to not only help in the development of immunological knowledge, but also bring new ideas from immunology into engineering in a more principled manner. This paper is concerned with the former: immunological modelling. The engineering implications have, to date, not been considered. Visceral leishmaniasis is a disease which, though it does not play a role in central and northern Europe, is still a deathly threat to many people around the world, especially in the poorest countries. The disease is caused by a parasite, that can hide within specific immune cells in the liver. Those infected cells of the immune system can attract more immune cells to the site of the infection. Many immune cells together can form a larger structure called “granuloma”. Our primary interest is in exactly how this granuloma can develop, and why some granulomas establish in a few days, whereas other sites of infection do not show any granuloma structures until weeks after the infection. Joining methods from computer science and complex sys- tems research with theories and experimental data from bi- ology promises to elevate our understanding of biology to a new level. Biology has been very successful in gathering huge amounts of data over the past decades. But only recently systems biology has emerged as a new trend within the discipline focusing on how the many facts can be pieced together to obtain a picture of the whole system. It is generally believed that models have an important place in this endeavor. Agent-based modelling techniques are one approach that can be used to incorporate known facts into a model [4]. This model can then be used to explore the dynamics and to test hypotheses about the complex system that is instantiated by the agents and their interactions. However, to construct a model, it is necessary to initially describe the domain knowledge. The Unified Modeling Language (UML) [5] seems to be a very suitable tool to describe biological systems [3]. In this study, UML was used to produce an initial conceptual model of the process of granuloma formation within the liver, before in a second step, developing the simulation model which is a implementation of the domain model as a computer simulation. In each step, special attention was paid to the assumptions and simplification that had to be undertaken. This is of vital importance in the modelling process [6]. II. AGENT BASED MODELLING AND THE I MMUNE SYSTEM The protection provided by the immune system emerges from the interaction of the millions of immune cells. There is no central control coordinating the behavior. Cytokines released by macrophages, for example, attract other cells to the area of infection. Attracted cells find their way towards the source of infection by sensing a local gradient of cytokine concentrations. The behavior of a cell in the immune system critically depends on the environment it is in. For example, a lymphocyte binding to an antigen will become activated in the presence of co-stimulatory signals, whereas in the absence of co-stimulation binding to an antigen might lead to cell death. Furthermore, the behavior of a cell might not only depend on the presence or absence of a signal but also on its strength. The dynamics of the immune system therefore also depend on the number of cells involved. By initiating an inflammatory response that attracts many immune cells, the immune system creates the environment that is needed for the single cell of the system to function properly. It is always tempting to assign capabilities of a system to a single agent within that system: “Macrophages ingest bacteria” or “T-cells kill infected cells”. However, in an isolated system containing just that agent and its target, it is likely that exactly this would not happen. For example, if we just had a test tube that contained some blood, in which macrophages and bacteria were present, it is conceivable that they would never meet. In the body, the chance for them to meet might be much