AGENT BASED MODELING (Background Info)
The seminal ideas for the science of “Complexity” originated decades ago with studies of social insects such as ants, bees and wasps. These insects have been in existence for approximately 100 million years, indicating that they must have some very successful strategies for survival. Those who studied them noted that although the individual insects were very simple creatures, they lived in complex societies with a clear division of labor. These societies have no centralized control, all of the individual insects follow simple internal rules for behavior based on information that they can perceive from their local environment. There is some degree of flexibility in task allocation, in cases such as shortages of food, colony members that have maintenance tasks join in the foraging effort. Though each individual is expendable, the colonies are robust, perturbations that kill some of the members do not bring an end to the colony. The insects also have the ability to communicate with each other indirectly, through the use of chemical signals called pheromones. Thus the colony acts as a network of autonomous peers each influencing one another. Though these social insects are simple creatures, they are capable of doing very complex things such as caring for young, building nests, foraging, and husbandry indicating that the observable complexity of these insect societies is greater than the sum of their simple parts.
When it was realized that many, if not all, natural systems share the properties described above for social insects, the study of natural systems as complex systems or complex adaptive systems was born. These properties, first known as “Swarm Intelligence”, have since been used to study and sometimes even predict the behavior of ecologies, societies, economies, biologies, business firms, organizations, financial markets, supermarkets, cities, schools of fish, flocking birds, traffic patterns, and terrorist networks. For instance take the example of a rush hour traffic jam. Each morning individuals get into their car and drive to work. While driving they are individually reacting to the cars around them, weaving in and out of lanes in order to get to work as fast as possible. No one gets in his or her car with the intention of creating a traffic jam. However, one occurs.
Investigators in other disciplines have noticed the parallels between the immune system and swarm intelligence, and for the past decade, the immune system has been known by those who study complexity to be a prime example of a complex system. Even though a small minority of immunologists are aware of this characterization of the immune system, some interesting work has already been published. One goal of our efforts will be for all students (and professors) of immunology (and biology) to learn about the complexity science view of the immune system and biological systems in general. This will potentiate a more complete understanding of immunology, and the brightest minds will be opened to explore the implications of these ideas for the enhancement of the treatment of diseases that involve the immune system.
Why Use Complexity to Study Biological Systems
Scientific research follows the basic principles of Analytic Thinking, meaning a system functions as a sum of its parts and can be understood by analyzing and dissecting its parts. In the laboratory, systems are narrowed down into tightly controlled systems to see how each individual entity works ignoring the complex nature of the system. This school of thought works for physics and chemistry where pieces make up the whole, but this premise may not be correct for biological systems. Using analytic thinking a large pool of observations and data accumulates and researchers use this vast amount of information to reconstruct the system based on their observations, leading to a skewed perception with a lot of details and little context. This leads to a poor understanding of the system in its entirety.
Biological systems, such as the immune system, are made up of a network of interactive elements. Therefore complexity should not be ignored, and systems thinking rather than analytical thinking should be applied to its pathways. A system is an integrated whole whose properties arise from the relationships between the parts not by the parts themselves. There is a shift from the importance of elements to the importance of organizational pattern. Look at the two paintings below.
Each is made using the same set of colors, but they are two completely different paintings and cannot be compared. A piece cannot be taken out of one and put in the other. It is the relationship of the colors, not the colors themselves that create the pattern. Therefore patterns of elements need to be identified and mapped, not just the elements themselves.
It is well known that the Immune system is a network of interacting pieces. Because of its complexity it can maintain coherence under change and is capable of learning and adapting. Leukocytes, also known as white blood cells, behave very much like ants and the immune system behaves very much like an aunt colony following colony rules of its own. In the immune system, each cell perceives what is in its environment via receptors on its surface that detects cytokines and chemokines (signals) displayed on the surface of other cells contacting them. Their perceptions are determined by the location in the body and the signals contacted. Their response affects their environment and therefore what the other immune cells see. Each one responds predictably to defined patterns of environmental signals. There is no centralized controlling entity. All leukocytes migrate, proliferate, and function based on their perceptions and are then reacting to the environmental signals produced by the other leukocytes around them. These swarm properties create larger actions which are unpredictable and outweigh the cellular components that constitute it. A full understanding of the immune system will never be obtained without understanding this complex nature (i.e. swarm functions) of leukocytes.
Agent Based Modeling:
The sort of agent-based modeling that we have used to create the Basic Immune Simulator is also known as “Individual-based modeling”, “Bottom-up modeling” or “Pattern-Oriented Modeling”, depending on the emphasis of the particular system being modeled and the preference of the modeler. Agent based modeling is a computational method used to study complex systems. In agent based modeling, agents represent active elements of the system in a computer simulated environment. The computer program for the model is written from the point of view of each of the agent types represented in the model. Behavior can be described in terms of activities or conditional (if-then) statements, in language that is natural to the expert in a field of study. Once the relevant agents and behavior are identified, rules for the behavior are formulated. The agents and behavior are then programmed into a computer simulation. Once the computer simulation is created and validated one may study the behavior of the system as a whole, looking for emergent properties in the observed behavior. One may change the rules for the behavior of the agents, to ask and answer “What if…?” sorts of questions about the behavior of a system.
There are many software libraries available for creating agent-based models, among the most popular are RePast, Swarm, and NetLogo. The differences between these open-source simulation toolkits include the programming languages used to implement them or the lack of a necessity for programming experience, as is the case with NetLogo. Agent-based modeling is a computational technique that is destined to come out of obscurity in the next few years, and we hope that our work will promote awareness of its utility.