| Popis: |
Classification is a major task in the gene sequence analysis. Based on the general principle of artificial immune system, this paper first constructed a classifier which inducted antibody-antigen identification, immune colonel reproduction, hypermutation, affinity mature and the network suppression, by simulating how the antigens stimulate the immune network and how the immune network responds. Then, a "leave-one-out" method was adopted to test the classifier's performance, applying 1-20th DNA sequences of Art-model-data with class attribute. Its accuracy was up to 90%. At last, a well-pleasing result was got on the prediction of 21-40th DNA sequences of Art-model-data. |
