摘要
利用分子的VolSurf参数预测化合物的水溶解度并阐明有利于水溶解度的主要分子结构特征.被测化合物包括185个共两大系列分子,使用偏最小二乘判别分析和多元线性回归方法在实验数据和分子特征之间建立相关性,均得到较好的结果.以70个化合物所建立的训练集模型对其余115个化合物有较好的预测能力.参数分析表明分子内较大的亲水区域对水溶解度有利;分子质心与疏水区、亲水区之间的不平衡性越高,水溶解度越大;分子量及体积大的分子对其水溶解度不利.
Water solubility of compounds was predicted and favorable molecular properties were determined with molecular VolSurf parameters. The tested molecules include two series among 185 compounds. Partial least squares (PLS) discriminant analysis and multiple linear regression (MLR) were used to correlate the experimental data with molecular properties and good results were obtained. The model from training set of 70 compounds had a good predictivity for the other 115 compounds. Descriptor analysis demonstrated that strong hydrophilic regions in the molecule were favorable to the solubility in water. High imbalance between the center of mass of a molecule and the barycentre of its hydrophilic and hydrophobic regions is beneficial to water solubility. High molecular weight and large volume are detrimental to solubility.
出处
《浙江大学学报(理学版)》
CAS
CSCD
2004年第6期661-666,共6页
Journal of Zhejiang University(Science Edition)
基金
宁波市博士基金资助项目(2004A610018).