摘要
为简化细胞病毒T细胞(cytotoxicity T lymphocytek,CTL)表位鉴定方法,应用改进的人工神经网络方法定量研究了短肽与MHC(major histocompatibility complex)分子结合亲合力的关系,并建立了CTL表位的预测模型,得到了预测模型最优性能参数.用此模型对短肽与HLA-A*0201分子结合的805个预测样本进行了预测,预测准确度达到73.8%.对来自黑色素MAGE-2的短肽与MHC分子的结合亲合力也进行了预测,结果较好.
To simplify the identification method of cytotoxicity T lymphocytek (CTL) epitopes, the binding affinity relationship between peptide and MHC (major histocompatibility complex) molecular was quantitatively studied with the modified artificial neural networks method. The model of predicting the epitopes of CTL was established, and the optimum parameters for the prediction model were obtained. The binding affinity prediction to the 805 peptides samples of HLA-A * 0201 molecular was carried on with this model, and the accuracy of prediction was 73. 8%. The affinity between nonamer of melanin MAGE-2 and MHC is predicted, and the results are reliable.
出处
《大连理工大学学报》
EI
CAS
CSCD
北大核心
2007年第4期473-478,共6页
Journal of Dalian University of Technology