Protein sequence and expression databases (transcriptomes) contain the information required to identify epitopes capable of generating protective immune responses in humans. A key event in the initiation of an immune response against disease is the presentation of antigenic peptide epitopes to T cells by human leukocyte antigen (HLA) molecules. Computational filtering tools that allow the prediction of HLA/epitope interaction can be applied to sequence databases to select for candidate epitopes, thus minimising the subsequent amount of laboratory work. Here, the basic principles of epitope prediction and a summary of the available prediction approaches are presented, with a particular emphasis on the use of algorithms based on virtual HLA-II quantitative matrices, capable of predicting promiscuous HLA-II ligands.
Read full abstract