论文部分内容阅读
肽的亲(疏)水性在生物活性肽的合理设计和药物发现中是非常重要的参数。基于表征氨基酸性质的HMLP参数,分别用直接替代法和加和替代法表征肽段的结构信息,采用留一法交叉检验,利用支持向量回归(SVR)构建了36个二肽辛醇-水分配系数(logP)的预测模型(ε=0.001,σ=7.5,C=200和ε=0.001,σ=10,C=200),结果表明加和替代法要好于直接替代法,且其留一法交叉检验的复相关系为0.8768,其预测结果与实验值一致。但二者都说明了logP主要与肽的分子表面积和亲水指标有关,同时说明HMLP参数能够较好地表征氨基酸或肽链的结构信息。
The pro-hydrophobic nature of peptides is a very important parameter in the rational design and discovery of bioactive peptides. Based on the HMLP parameters that characterize the nature of amino acids, the structure information of the peptides was characterized by direct substitution method and additive substitution method respectively. One-way crossover test and support vector regression (SVR) were used to construct 36 dipeptide octanol-water partition (Ε = 0.001, σ = 7.5, C = 200 and ε = 0.001, σ = 10, C = 200). The results show that the addition and substitution method is better than the direct substitution method, Cross-test of the complex correlation of 0.8768, the predicted results consistent with the experimental value. However, both of them indicate that logP is mainly related to the molecular surface area of the peptide and the hydrophilic index, and at the same time shows that the HMLP parameter can better characterize the structure information of amino acids or peptide chains.