Antibody Immunogenicity Prediction

What is an antibody?

Antibody is a kind of immunoglobulin formed by the body under the stimulation of antigen, which can specifically bind to the antigen, participates in neutralizing toxins, sterilizing and lysing. Antibody production refers to a process in which an organism produces an immunoglobulin that is specifically reacted with a corresponding antigen by a plasma cell differentiated by B cells under the stimulation of an antigenic substance.


Since electrophoresis was first used to demonstrate that the antibody activity in serum is in the gamma globulin portion, the antibodies have been collectively referred to as two (gamma) globulins. It was later proved that the antibodies were not all in the γ region; and the globulins located in the γ region did not necessarily have antibody activity. In 1964, the World Health Organization held a special meeting to refer to antibody-related globulins as immunoglobulins (Ig). Such as myeloma protein, macroglobulinemia, cryoglobulinemia and other abnormal immunoglobulins present in the serum and “normal human” naturally occurring immunoglobulin subunits. Thus immunoglobulins are the concept of structure and chemistry, and antibodies are biological and functional concepts. It can be said that all antibodies are immunoglobulins, but not all immunoglobulins are antibodies.


Antigen has two basic properties, namely antigenicity and immunogenicity.

  1. Antigenicity refers to the ability of an antigen to specifically bind to an antibody or sensitized lymphocyte induced by it. The strength of antigenicity is closely related to the size of the antigen molecule, the chemical composition, the structure of the antigenic determinant, and the proximity of the antigen to the immune animal. It is generally believed that the larger the molecular weight of the antigen, the more complex the chemical composition, the more complete the three-dimensional structure and the farther the relationship with the immunized animal, the stronger the antigenicity will be.


  1. Antibody immunogenicitymeans the ability to stimulate the body to form specific antibodies or sensitized lymphocytes. That is to say, the antigen can stimulate specific immune cells, activate, proliferate, and differentiate immune cells, and finally produce antibodies to immune effector antibodies and sensitized lymphocytes. It also refers to the specific immune response of the body’s immune system to form antibodies or sensitized T lymphocytes after the antigen stimulates the body.


T cell epitope refers to an epitope recognized by a TCR, and the epitope component is a polypeptide after protein degradation. Such epitopes are generally not located on the surface of the antigen molecule, and the antigen-presenting cells must process the antigen into a small molecule polypeptide and bind it to the MHC molecule before being recognized by the TCR. Therefore, studying T cell epitopes not only helps to understand the immune response mechanisms of infectious diseases, autoimmune diseases, allergic reactions and tumors, but also promotes the design of computer-aided vaccines. The rapid development of bioinformatics provides an effective way to study T cell epitopes.


In view of the current research status of T cell epitopes, a literature has selected several hot issues to carry out antibody immunogenicity prediction research. Main tasks are as follows:


  • Combining sequence information and structural information to predict antigenic epitopes is the development direction of antigen epitope prediction. The HLA-A2 molecule is a kind of MHC class molecule ubiquitous in the human population. The HLA-A2 antigen peptide was selected as the research object, and the amino acid physicochemical properties of the antigen peptide and the energy term when combined with MHC were explored. In order to avoid the redundancy of the features used, affecting the efficiency and performance of the classification, the features were screened by the method of maximum correlation minimum redundancy (MRMR), and finally 50 features with high recognition ability were selected and constructed. A prediction method for HLA-A2 class molecular antigen peptides. The predicted results indicate that the selected features are capable of efficiently identifying antigen peptides bound by HLA-A2 molecules.


  • The I-Ag7 molecule and the HLA-DQ8 molecule are MHC class II molecules associated with type I diabetes in mice and humans. To predict the restriction epitopes of these two classes of MHC molecules, a GPS-MBA software package was developed. With improved Gibbs sampling algorithm, the core sequence of the epitope was obtained that were scored using GPS algorithm to construct a prediction model. After a comprehensive evaluation and comparison, GPS-MBA performed better than the same type of software. A large number of potential I-Ag7 and HLA-DQ8 epitopes can be predicted using this software. In addition, the Epitope Database for Type I Diabetes (TEDB) was designed, which contains all experimentally validated or predicted epitopes associated with type I diabetes.


  • In order to verify the negative selection hypothesis and study the immunogenicity of the antigenic epitope, the exogenous T cell epitope and the non-epitope peptide and the host protein sequence were analyzed for both human and mouse hosts, respectively. Using sequence alignment, it was found that only a very small number of epitopes were similar to the host protein sequence, which was lower than the number of non-epitope peptides similar to the host protein sequence. This result indicates that the exogenous T cell epitope is similar to the host protein sequence, suggesting that using an epitope with low similarity to the host protein sequence can help reduce the probability of cross-immunization and enhance the vaccine immunogen.


(4) In order to further develop the proteasome digestion site prediction software to improve the accuracy of predicting antigenic epitopes, an evaluation of existing prediction software has been carried out adopting a number of methods and software to predict proteasome cleavage sites. The more commonly used computer programs are PAProC, MAPPP and NetChop. Performance evaluation of the three using a unified data set shows that NetChop performs better than the other.