Most epitopes recognized by antibodies or B cells can be thought of as three-dimensional surface features of an antigen molecule; these features fit precisely and thus bind to antibodies. Exceptions are linear epitopes, which are determined by the amino acid sequence (the primary structure) rather than by the 3D shape (tertiary structure) of a protein.
T cell epitopes are presented on the surface of an antigen-presenting cell, where they are bound to MHC molecules. T cell epitopes presented by MHC class I molecules are typically peptides between 8 and 11 amino acids in lengths, whereas MHC class II molecules present longer peptides, and non-classical MHC molecules also present non-peptidic epitopes such as glycolipids.
Genetic sequences coding for epitopes that are recognised by common antibodies can be fused to genes, thus aiding further molecular characterization of the gene product. Common epitopes used for this purpose are c-myc, HA, FLAG, V5.
Epitopes are sometimes cross-reactive. This property is exploited by the immune system in regulation by anti-idiotypic antibodies (originally proposed by Nobel laureate Niels Kaj Jerne). If an antibody binds to an antigen's epitope, the paratope could become the epitope for another antibody that will then bind to it. If this second antibody is of IgM class, its binding can upregulate the immune response; if the second antibody is of IgG class, its binding can downregulate the immune response.
Intensive research is currently taking place to design reliable tools that will predict epitopes on proteins.