Major Histo-Compatibility (MHC) class I molecules are major agents of the mammalian adaptive immune system. Class I molecules bind short antigenic peptides with a length of 8-10 residues in the Endoplasmatic Reticulum (ER) and after transport to the cell surface the peptides are presented to T-lymphocytes. The binding site of class I molecules is formed by a deep cleft between two α-helices at top of an extended β-sheet. Only tightly bound high-affinity peptides have a chance to reach the cell surface and trigger an immune response. It is therefore of great interest to identify possible high-affinity antigenic peptides that could be used as vaccines to help the immune system to detect viral infections or kill malignant cells. A large number of crystal structures of antigenic peptides in complex with class I alleles have been determined that allow to understand the structural details important for peptide binding. Biophysical and biochemical analysis of peptide-class I complexes has resulted in a number of rules concerning the selection of high-affinity peptides. However, an accurate prediction of allele specific peptide-binding is still not possible. This issue is currently addressed by various computational tools developed by the bioinformatics community. The computational efforts range from statistical analysis of peptide motifs stored in databases to application of neural network methods and support vector machine approaches. In addition, structure based approaches to predict class I binding specificity including molecular modeling and molecular dynamics (MD) simulations will also be presented.
Keywords: Class I molecule, antigen presentation, binding affinity prediction, machine learning, ligand receptor docking, neural network prediction