摘要
目的研究了CTL表位与MHC分子结合的定量构效关系模型。方法采用偏最小二乘法(PLS)对校正集进行校正,并对预测集中的样本进行预测。结果得到比较满意的短肽-MHC分子结合亲合力与短肽氨基酸序列之间的关系式;模型的预测精确度达到66.8%。结论偏最小二乘方法可作为CTL表位预测方法。用该模型对预测样本进行短肽-MHC分子的结合亲合力的预测,结果令人满意,说明这种模型在预测CTL表位方面是可行的。
Objective To study the model of the quantitative structure-activity relationship (QSAR) CTL epitopes binding to MHC molecule. Methods Partial least square (PLS) raethod was used to train the training sets and predict the predictive sets. Results Asatisfied formula between the binding affinity of peptide-MHC complex molecule and the sequence of peptide was obtained. The predictive accuracy of the model was 66.8 %. Conlusion Partial least squares can be used as a raethod for predicting CTL epitopes.
出处
《免疫学杂志》
CAS
CSCD
北大核心
2007年第2期166-171,共6页
Immunological Journal
关键词
CTL表位
定量构效关系
偏最小二乘法
CrL epitopes
Quantitative structure-activity relationship
Partial least square