The major assignment of an immune system is to defend the host against infections, a task which clearly is essential for any organism. While surprisingly many other organismal traits may be linked to individual genes, immune systems have always been viewed as systems, in the sense their genetic foundation is complex and based on a multitude of proteins in many pathways, which interact with each other to coordinate the defense against infection.
Full scale computational models for the entire immune system is therefore also not going to be simple, but will rely on integration of many different components. However, many of these components may be much simpler models of how immune systems --- step by step --- deal with pathogenic organisms.
Thus, the optimal way of studying immune systems would be to carry out analysis at several levels including comparative genomics and proteomics, co-evolution with pathogens, tissue specific processes, regulation networks, population dynamics, etc. In other words, today studies of immune systems calls for a multi-disciplinary approach, where bioinformatics, genomics, proteomics, cellular, molecular and clinical immunology and mathematical modeling can in combination provide efficient answers to many of the basic problems in immunology. In recent years several success stories (especially within HIV research) have demonstrated the necessity of such a multi-disciplinary approach.
Two closely related research fields are trying to follow this multi-disciplinary approach to study immune systems:
We have recently published the first book on immunological bioinformatics,
where we define this research field as the research field that applies
informatics techniques to generate a systems-level view of the immune
system. Contrary to theoretical immunology, immunological bioinformatics is
a new field, and its main focus has been (so far) to get better insights into
the specificity of the molecules involved in antigen presentation and
processing pathways. The immune system does not react to entire pathogens but
rather to parts of these called epitopes. A major aim of this research field is
to develop methods that can be used to identify epitopes in basicly any genome.
The predicted epitopes can eventually be used in designing a vaccine. Besides
developing several epitope prediction methods (see
Servers ), we are establishing a research line in Utrecht that
investigates the evolution of antigen processing and presentation.
Antigen processing. Our main contribution to immunological bioinformatics has been to increase the understanding of the intracellular protein degradation. Our novel approach to predict the specificity of the proteasome Kesmir, et al, 2002 ; Nielsen, et al, 2005 was demonstrated to be superior to the classical approaches Saxova et al, 2003 . The method is available \via the internet, NetChop, and processes approximately 200 protein degradation queries every day. Together with Colin Watts (Dundee) we are now developing methods that deal with intra-vesicular degradation within a cell, which provides ligands for MHC class II presentation (see eg, AEP predictor ).
There are two forms of proteasome in most cells: the constitutive proteasome which is expressed by all cells and the immunoproteasome, which is expressed by infected cells, or cells that are stimulated by cytokines. We have shown that the immunoproteasome is a more specific enzyme than the constitutive proteasome ( Kesmir, et al, 2003). The pathogens use this fact and escape the degradation by immunoproteasome, but not by the constitutive proteasome ( Yusim et al, 2002). The rate of proteasomal cleavage depends on the length of the substrate and follows Michaels-Menten kinetics, and the distribution of fragments depends on the gate size ( Luciani et al., 2005 ).
Antigen presentation. The MHC-peptide interaction is the oldest research area within immunological bioinformatics. However, the methods developed focused on a few MHC alleles, which limited the study of the specificity of the immune response to a small fraction of the human population. Together with our experimental collaborators, we enlarged these predictions to cover almost 99% of the human population by developing NetMHC method ( Buus et al, 2003). We have established a solid method to classify MHC binding specificities into MHC supertypes ( Lund et al 2004 ), and developed mathematical models to explore factors generating MHC polymorphism ( De Boer et al, 2004). More interesting theoretical work on MHC polymorphism can be found in Utrecht theoretical immunology group .
Evolution of antigen processing and presentation. We have shown that the specificity of human MHC molecules has evolved to fit the specificity of the immunoproteasome ( Kesmir, et al, 2003). Thus good MHC ligands also have a high probability to be generated by the immunoproteasome. TAP is adapted even more to the specificty of human MHC molecules ( Nielsen, et al, 2005 ). In this way the efficacy of the antigen processing and the presentation seemed to be optimized.
In collaboration with Nigel Burroughs we have enumerated all class I MHC peptides in the human proteome and in various pathogens to determine the likelihoods that immunodominant peptides overlap between self and non-self. Due to the apparently high information content of the 9-mers used in class I antigen presentation the overlaps are very small (Burroughs et al., 2004). This analysis also confirms the co-evolution of the specificities of proteasome, TAP, and MHC molecules.