Abstract: The goal of this study is to develop an automated computerized method for the identification of helper T-cell epitope. In order to describe the peptide structure, a new set of descriptors called sequence property vector(ξ) was developed from the primary structures of peptides based on the distance between amino acid side chains and on the electric property of each amino acid side chain, and looking on each amino acid side chain as a pseudo atom. Then a quantitative model was proposed for 28 immunogenic pep-tides including 14 A^dand non-A^drestricted peptide sequence, assigned active and inactive peptides, respectively. In order to explain the accuracy of the prediction system, random sampling validation and cross validation were performed. The results showed that the prediction system exhibited a good stability and excellent prediction ability. This method may also be applied in the study to predict the MHC classes Iand Iepitope of human, and it should be useful in immune recognition studies of protein antigens and the design and study of subunit vaccine.

Key words: Epitope prediction, MHC restriction, Topological characteristic, Sequence property vector