CONSIDERATIONS
T he immune system functions to protect us from disease-causing organisms. In order to perform this function the immune system must be able to recognize a potentially large variety of pathogens. Recognition in the immune system is performed on the basis of chemistry, and thus the immune system, rather than recognizing whole organisms, recognizes foreign molecules, or antigens as they are called by immunologists. The number of possible antigens is extremely large, and the immune system has only finite resources to devote to antigen recognition. Thus, devising efficient strategies for antigen recognition has been essential in the evolution of a functional immune system.
Our goal in this article is to elucidate some design principles for an immune system that has to cope with recognizing a large number of antigens. Because pathogens are living organisms, they too evolve. Thus, the task of the immune system is to recognize pathogens even if they evolve into somewhat changed forms. The signature of evolution in a pathogen is the change in some molecules; that is, antigens can change in time due to mutation.
Here we consider two problems, one in which the antigens remain fixed during the time of the immune response, and one in which the antigens change rapidly enough for that change to occur during an immune response. Space does not permit a thorough discussion of the multitude of facts and existing theoretical considerations about the immune system. For an introduction to some of the material, we refer the reader to a recent review .
Immune recognition involves a type of pattern recognition, and as such the principles that we discuss have application to problems outside the realm of immunology. For example, some of the ideas that we present have found application in computer security [2] and shape spaces, which are a key concept in the article, appear also in morphometry [3] and computer vision [4].